Enterprise digital intelligent decision-making system
Through the enterprise's digital intelligent decision-making system, project allocation is optimized using prediction models and genetic algorithms, the problems of delays and resource idleness caused by uneven team tasks are solved, and the project is completed on time and resource utilization is efficient.
Patent Information
- Application Number
- CN202411089391.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-09
- Publication Date
- 2024-12-03
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In large enterprise environments, improper project allocation leads to excessive delays or insufficient tasks in some teams, resulting in idle resources. The existing technology relies on subjective judgment and lacks effective solutions.
A digital intelligent decision-making system for enterprises is designed to collect project and team data through data acquisition modules, build prediction models and generate scheduling schemes in combination with genetic algorithms to ensure reasonable allocation of projects to avoid delays and idle resources.
Through the prediction model, estimate the cost and time of the team completing projects, generate the optimal scheduling plan, ensure that the team completes the project on time and reduces idle resources.
Smart Images

Figure CN119067468A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent decision-making technology, and more specifically, to an enterprise digital intelligent decision-making system. Background Art
[0002] In a large enterprise environment, there is often the challenge of advancing multiple projects in parallel. How to efficiently and fairly allocate these projects to different teams has become an urgent problem to be solved. The current practice mostly relies on the subjective judgment of senior managers and self-recommendation of teams to complete project allocation.
[0003] However, the above method has the following defects: 1. Some teams may find it difficult to complete the project on time due to excessive workload, resulting in project delays; 2. Some teams may complete the task prematurely due to insufficient workload, resulting in idle team resources. Summary of the invention
[0004] The present invention provides an enterprise digital intelligent decision-making system to solve the technical problems in related technologies that a team is unable to complete a project on time due to excessive workload, resulting in project delays, and that team resources are idle due to insufficient workload.
[0005] The present invention provides an enterprise digital intelligent decision-making system, comprising:
[0006] A data collection module, which is used to collect project data of M projects to be assigned and team data of N teams of the enterprise;
[0007] The project data is represented by the first characteristic sequence, which includes U sequence units, and the u-th sequence unit represents the cost, the number of personnel required, the number of equipment required, and the project development cycle of the u-th sub-project, where 1≤u≤U;
[0008] The team data is represented by the second feature sequence, which includes V sequence units. The vth sequence unit represents the education, work experience, professional skills, and health status of the vth person, where 1≤v≤V;
[0009] A prediction model building module, which is used to build and train a prediction model based on the project data of the project to be assigned and the team data of the team;
[0010] The input of the prediction model is the project data of a project to be assigned and the team data of a team, and the output values represent the estimated cost and estimated completion time respectively;
[0011] A scheduling scheme generation module, which is used to generate a scheduling scheme based on a prediction model combined with a genetic algorithm;
[0012] According to the scheduling plan, the company's M unassigned projects are assigned to N teams.
[0013] Furthermore, M and N are both custom parameters, M is set according to the number of projects to be assigned to the enterprise, and N is set according to the number of teams.
[0014] Furthermore, educational background includes: junior high school, high school and college, and educational background is represented by real number codes; work experience is represented by years of work; professional skills are represented by skill scores, and skill scores are obtained through questionnaire tests; health status includes: healthy and unhealthy, and is represented by real number codes.
[0015] Furthermore, preprocessing is performed on the project data of the M projects to be assigned and the team data of the N teams, including the following steps:
[0016] Step S201, the missing values in the project data and the team data are filled by taking the average value of the previous sequence unit and the next sequence unit;
[0017] Step S202, replacing the outliers in the project data and team data by calculating the mean of all non-outliers;
[0018] Step S203, normalizing the project data and team data using the maximum and minimum normalization method.
[0019] Further, the prediction model includes: a first hidden layer, an extraction layer, a second hidden layer, a third hidden layer, a first classifier, and a second classifier;
[0020] The first hidden layer includes U hidden units, the u-th hidden unit inputs the u-th sequence unit of the first feature sequence, and outputs the first hidden feature;
[0021] The extraction layer is used to extract the first hidden feature output by the U-th hidden unit as the first update feature;
[0022] The second hidden layer inputs the second feature sequence and outputs the second updated feature;
[0023] The third hidden layer is used to expand the first updated features output by the first hidden layer and the second updated features output by the second hidden layer into vector representations and concatenate them to obtain a combined vector;
[0024] The combined vector is input to the first classifier and the second classifier, the classification space of the first classifier represents the estimated cost, and the classification space of the second classifier represents the estimated completion time.
[0025] Furthermore, the calculation formula for the u-th hidden unit of the first hidden layer includes:
[0026]
[0027] where x urepresents the u-th sequence unit of the first feature sequence input to the u-th hidden unit, h u and h u-1 denote the first hidden features output by the u-th and u-1-th hidden units, respectively, and h 0 Assign a value of 0, r u 、z u and denote the reset gate, update gate, and candidate hidden feature of the u-th hidden unit in the first hidden layer, respectively. and They respectively represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the reset gate of the u-th hidden unit of the first hidden layer, and They respectively represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the reset gate of the u-th hidden unit of the first hidden layer, and They respectively represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the candidate hidden features of the u-th hidden unit of the first hidden layer, ⊙ represents point-by-point multiplication, sigmoid represents the sigmoid activation function, and tanh represents the tanh activation function.
[0028] Furthermore, the calculation formula of the second hidden layer is as follows:
[0029]
[0030] Where Attention represents the second updated feature output by the second hidden layer, Query, Key and Value represent the first intermediate feature, the second intermediate feature and the third intermediate feature of the second hidden layer respectively, T represents the transposition operation, and d k Represents the number of dimensions of each sequence unit of the second feature sequence, which is a constant value of 4, and softmax represents the softmax activation function.
[0031] Furthermore, the loss function of the prediction model adopts the cross entropy loss function.
[0032] Furthermore, a scheduling scheme is generated based on the prediction model combined with a genetic algorithm, including the following steps:
[0033] Step S301, generating an initialization population with Q number of individuals meeting the constraint conditions, where Q is a custom parameter;
[0034] Constraints include: one unassigned project can only be assigned to one team; one team can only be assigned up to three unassigned projects;
[0035] The individuals of the initialized population are represented by matrix coding, which is expressed as:
[0036]
[0037] in They respectively represent the allocation of the 1st to-be-allocated project to the Mth to-be-allocated project to the 1st team. They respectively represent the allocation of the first project to be allocated to the first team to the Nth team. Indicates the allocation of the Mth project to be allocated to the Nth team. The element value of the matrix code is represented by 0 or 1;
[0038] Step S302, constructing an objective function through the prediction model, and obtaining the fitness value of each individual in the initialized population through the objective function calculation;
[0039] Step S303, selecting two individuals with the largest fitness values in the initialization population as parents;
[0040] Step S304, generating a random number in the range of 0 to 1, and when the random number is less than a first threshold, performing a crossover operation that satisfies the constraint condition;
[0041] The first threshold is a custom parameter;
[0042] The crossover operation means randomly selecting the element value of one position in the matrix encoding of two individuals in the parent generation to exchange;
[0043] Step S305, generating a random number in the range of 0 to 1, and when the random number is less than a second threshold, performing a mutation operation that satisfies the constraint condition;
[0044] The first threshold is a custom parameter;
[0045] The mutation operation means randomly selecting an individual in the parent generation to regenerate the element value of one position in the matrix encoding;
[0046] Step S306, if it is determined that the current number of iterations is greater than or equal to the maximum number of iterations, the individual with the largest fitness value in the initialized population is used as the scheduling solution, otherwise, the process returns to step S302, where the maximum number of iterations is a custom parameter.
[0047] Furthermore, the objective function is calculated as follows:
[0048]
[0049] Where Fit represents the fitness value of the individual, and W represents the estimated cost and estimated completion time of the mth project to be assigned and the nth team’s team data respectively. 1and W 2 Respectively represent the first weight parameter and the second weight parameter, W 1 and W 2 All are custom parameters and W 1 and W 2 The sum of is 1.
[0050] The beneficial effect of the present invention is that the present invention estimates the cost and completion time of different teams responsible for different projects through a prediction model, and generates an optimal scheduling plan based on the prediction model combined with a genetic algorithm to ensure that the team can complete the project on time and reduce the idleness of team resources. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] Figure 1 It is a schematic diagram of an enterprise digital intelligent decision-making system of the present invention;
[0052] Figure 2 It is a flow chart of preprocessing project data and team data of the present invention;
[0053] Figure 3 It is a flow chart of the prediction model of the present invention combined with the genetic algorithm to generate a scheduling plan;
[0054] In the figure: data collection module 101, prediction model building module 102, scheduling plan generation module 103. DETAILED DESCRIPTION
[0055] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.
[0056] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.
[0057] like Figure 1 to Figure 3 As shown, an enterprise digital intelligent decision-making system includes:
[0058] The data collection module 101 is used to collect project data of M projects to be assigned and team data of N teams of the enterprise;
[0059] The project data is represented by the first feature sequence, which includes U sequence units, and the u-th sequence unit represents the cost, the number of personnel required, the number of equipment required, and the project development cycle of the u-th sub-project, 1≤u≤U;
[0060] The team data is represented by the second feature sequence, which includes V sequence units. The vth sequence unit represents the education, work experience, professional skills, and health status of the vth person, 1≤v≤V;
[0061] A prediction model building module 102, which is used to build and train a prediction model based on the project data of the project to be assigned and the team data of the team;
[0062] The input of the prediction model is the project data of a project to be assigned and the team data of a team, and the output values represent the estimated cost and estimated completion time respectively;
[0063] A scheduling scheme generating module 103, which is used to generate a scheduling scheme based on a prediction model combined with a genetic algorithm;
[0064] According to the scheduling plan, the company's M unassigned projects are assigned to N teams.
[0065] In one embodiment of the present invention, both M and N are custom parameters, M is set according to the number of projects to be assigned by the enterprise, and N is set according to the number of teams.
[0066] In one embodiment of the present invention, the enterprise's project data estimation can be carried out through a survey of similar historical projects or subjective opinions. For example, when estimating the number of equipment required, a specific and reasonable estimation result can be obtained based on the data of the number of equipment required for similar historical projects; the enterprise's team data can be based on the actual situation of employees, skill test results, or subjective evaluation based on experience. For example, when collecting professional skill data, employees can be scored based on their actual performance in historical projects and the results of possible skill tests.
[0067] In one embodiment of the present invention, the educational background includes: junior high school, high school and university, and the educational background is represented by a real number code, for example, the educational background of junior high school is represented by the real number 1.1; the work experience is represented by the number of years of work; the professional skills are represented by the skill score, and the skill score is obtained through a questionnaire test; the health status includes: healthy and unhealthy, and is represented by a real number code, for example, the health status of unhealthy is represented by the real number 0.
[0068] In one embodiment of the present invention, Figure 2 As shown, preprocessing is performed on the project data of M projects to be assigned and the team data of N teams, including the following steps:
[0069] Step S201, the missing values in the project data and the team data are filled by taking the average value of the previous sequence unit and the next sequence unit;
[0070] Step S202, replacing the outliers in the project data and team data by calculating the mean of all non-outliers;
[0071] Step S203, normalizing the project data and team data using the maximum and minimum normalization method.
[0072] In one embodiment of the present invention, the prediction model includes: a first hidden layer, an extraction layer, a second hidden layer, a third hidden layer, a first classifier, and a second classifier;
[0073] The first hidden layer includes U hidden units, the u-th hidden unit inputs the u-th sequence unit of the first feature sequence, and outputs the first hidden feature;
[0074] The extraction layer is used to extract the first hidden feature output by the U-th hidden unit as the first update feature;
[0075] The second hidden layer inputs the second feature sequence and outputs the second updated feature;
[0076] The third hidden layer is used to expand the first updated features output by the first hidden layer and the second updated features output by the second hidden layer into vector representations and concatenate them to obtain a combined vector;
[0077] The combined vector is input to the first classifier and the second classifier, the classification space of the first classifier represents the estimated cost, and the classification space of the second classifier represents the estimated completion time.
[0078] In one embodiment of the present invention, the calculation formula of the u-th hidden unit of the first hidden layer includes:
[0079]
[0080] where x u represents the u-th sequence unit of the first feature sequence input to the u-th hidden unit, h u and h u-1 denote the first hidden features output by the u-th and u-1-th hidden units, respectively, and h 0 Assign a value of 0, r u 、z u and denote the reset gate, update gate and candidate hidden feature of the u-th hidden unit in the first hidden layer, respectively. and They respectively represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the reset gate of the u-th hidden unit of the first hidden layer, and They respectively represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the reset gate of the u-th hidden unit of the first hidden layer, and They respectively represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the candidate hidden features of the u-th hidden unit of the first hidden layer, ⊙ represents point-by-point multiplication, sigmoid represents the sigmoid activation function, and tanh represents the tanh activation function.
[0081] In one embodiment of the present invention, the calculation formula of the second hidden layer is as follows:
[0082]
[0083] Where Attention represents the second updated feature output by the second hidden layer, Query, Key and Value represent the first intermediate feature, the second intermediate feature and the third intermediate feature of the second hidden layer respectively, T represents the transposition operation, and d k Represents the number of dimensions of each sequence unit of the second feature sequence, which is a constant value of 4, and softmax represents the softmax activation function.
[0084] In one embodiment of the present invention, the loss function of the prediction model adopts a cross entropy loss function.
[0085] In one embodiment of the present invention, Figure 3 As shown, the scheduling scheme is generated based on the prediction model combined with the genetic algorithm, including the following steps:
[0086] Step S301, generating an initialization population with Q individuals meeting the constraint conditions, where Q is a custom parameter, preferably, Q is set to 20;
[0087] Constraints include: one unassigned project can only be assigned to one team; one team can only be assigned up to three unassigned projects;
[0088] The individuals of the initialized population are represented by matrix coding, which is expressed as:
[0089]
[0090] in They respectively represent the allocation of the 1st to-be-allocated project to the Mth to-be-allocated project to the 1st team. They respectively represent the allocation of the first project to be allocated to the first team to the Nth team. Indicates the allocation of the Mth project to be allocated to the Nth team. The element value of the matrix code is represented by 0 or 1;
[0091] For example The value assigned is 0, which means that the first item to be assigned is not assigned to the second team;
[0092] Step S302, constructing an objective function through the prediction model, and obtaining the fitness value of each individual in the initialized population through the objective function calculation;
[0093] Step S303, selecting two individuals with the largest fitness values in the initialization population as parents;
[0094] Step S304, generating a random number in the range of 0 to 1, and when the random number is less than a first threshold, performing a crossover operation that satisfies the constraint condition;
[0095] The first threshold is a custom parameter. Preferably, the first threshold is set to 0.7;
[0096] The crossover operation means randomly selecting the element value of one position in the matrix encoding of two individuals in the parent generation to exchange;
[0097] For example, an individual in the parent generation corresponds to the matrix encoding The matrix encoding corresponding to another individual of the parent generation Make an exchange;
[0098] Step S305, generating a random number in the range of 0 to 1, and when the random number is less than a second threshold, performing a mutation operation that satisfies the constraint condition;
[0099] The first threshold is a custom parameter. Preferably, the first threshold is set to 0.05;
[0100] The mutation operation means randomly selecting an individual in the parent generation to regenerate the element value of one position in the matrix encoding;
[0101] For example, an individual in the parent generation corresponds to the matrix encoding is equal to 0, and the value after mutation is 1;
[0102] Step S306, if it is determined that the current number of iterations is greater than or equal to the maximum number of iterations, the individual with the largest fitness value in the initialized population is used as the scheduling scheme, otherwise return to step S302, where the maximum number of iterations is a custom parameter, preferably, the maximum number of iterations is set to 200.
[0103] In one embodiment of the present invention, the calculation formula of the objective function is as follows:
[0104]
[0105] Where Fit represents the fitness value of the individual, and W represents the estimated cost and estimated completion time of the mth project to be assigned and the nth team’s team data respectively. 1 and W 2 Respectively represent the first weight parameter and the second weight parameter, W 1 and W 2 All are custom parameters and W 1 and W 2 The sum of is 1.
[0106] The above describes an embodiment of the present embodiment, but the present embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, all of which are within the protection of the present embodiment.
Claims
1. An enterprise digital intelligent decision-making system, characterized in that: include: A data collection module, which is used to collect project data of M projects to be assigned and team data of N teams of the enterprise; The project data is represented by the first characteristic sequence, which includes U sequence units, and the u-th sequence unit represents the cost, the number of personnel required, the number of equipment required, and the project development cycle of the u-th sub-project, where 1≤u≤U; The team data is represented by the second feature sequence, which includes V sequence units. The vth sequence unit represents the education, work experience, professional skills, and health status of the vth person, where 1≤v≤V; A prediction model building module, which is used to build and train a prediction model based on the project data of the project to be assigned and the team data of the team; The input of the prediction model is the project data of a project to be assigned and the team data of a team, and the output values represent the estimated cost and estimated completion time respectively; A scheduling scheme generation module, which is used to generate a scheduling scheme based on a prediction model combined with a genetic algorithm; According to the scheduling plan, the company's M unassigned projects are assigned to N teams.
2. According to claim 1, the enterprise digital intelligent decision-making system is characterized in that: Both M and N are custom parameters. M is set according to the number of projects to be assigned by the enterprise, and N is set according to the number of teams.
3. According to claim 1, the enterprise digital intelligent decision-making system is characterized in that: Educational qualifications include: Junior high school, high school and college, academic qualifications are represented by real number codes; Work experience is expressed by years of work experience; Professional skills are expressed through skill scores, which are obtained through questionnaire tests; Health conditions include: healthy and unhealthy, which are represented by real number coding.
4. According to claim 1, the enterprise digital intelligent decision-making system is characterized in that: Preprocessing the project data of the M projects to be assigned and the team data of the N teams includes the following steps: Step S201, the missing values in the project data and the team data are filled by taking the average value of the previous sequence unit and the next sequence unit; Step S202, replacing the outliers in the project data and team data by calculating the mean of all non-outliers; Step S203, normalizing the project data and team data using the maximum and minimum normalization method.
5. According to claim 1, the enterprise digital intelligent decision-making system is characterized in that: The prediction model includes: a first hidden layer, an extraction layer, a second hidden layer, a third hidden layer, a first classifier and a second classifier; The first hidden layer includes U hidden units, the u-th hidden unit inputs the u-th sequence unit of the first feature sequence, and outputs the first hidden feature; The extraction layer is used to extract the first hidden feature output by the U-th hidden unit as the first update feature; The second hidden layer inputs the second feature sequence and outputs the second updated feature; The third hidden layer is used to expand the first updated features output by the first hidden layer and the second updated features output by the second hidden layer into vector representations and concatenate them to obtain a combined vector; The combined vector is input to the first classifier and the second classifier, the classification space of the first classifier represents the estimated cost, and the classification space of the second classifier represents the estimated completion time.
6. The enterprise digital intelligent decision-making system according to claim 5, characterized in that: The calculation formula for the uth hidden unit in the first hidden layer includes: where x u represents the u-th sequence unit of the first feature sequence input to the u-th hidden unit, h u and h u-1 They represent the first hidden features output by the u-th and u-1-th hidden units, respectively. h0 is assigned to 0, r u 、z u and denote the reset gate, update gate, and candidate hidden feature of the u-th hidden unit in the first hidden layer, respectively. and They respectively represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the reset gate of the u-th hidden unit of the first hidden layer, and They respectively represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the reset gate of the u-th hidden unit of the first hidden layer, and They respectively represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the candidate hidden features of the u-th hidden unit of the first hidden layer, ⊙ represents point-by-point multiplication, sigmoid represents the sigmoid activation function, and tanh represents the tanh activation function.
7. The enterprise digital intelligent decision-making system according to claim 5 is characterized in that: The calculation formula for the second hidden layer is as follows: Where Attention represents the second updated feature output by the second hidden layer, Query, Key and Value represent the first intermediate feature, the second intermediate feature and the third intermediate feature of the second hidden layer respectively, T represents the transposition operation, and d k Represents the number of dimensions of each sequence unit of the second feature sequence, which is a constant value of 4, and softmax represents the softmax activation function.
8. The enterprise digital intelligent decision-making system according to claim 1 is characterized in that: The loss function of the prediction model adopts the cross entropy loss function.
9. The enterprise digital intelligent decision-making system according to claim 1, characterized in that: The scheduling scheme is generated based on the prediction model combined with the genetic algorithm, including the following steps: Step S301, generating an initialization population with Q number of individuals meeting the constraint conditions, where Q is a custom parameter; Constraints include: one unassigned project can only be assigned to one team; one team can only be assigned up to three unassigned projects; The individuals of the initialized population are represented by matrix coding, which is expressed as: in They respectively represent the allocation of the 1st to-be-allocated project to the Mth to-be-allocated project to the 1st team. They respectively represent the allocation of the first project to be allocated to the first team to the Nth team. Indicates the allocation of the Mth project to be allocated to the Nth team. The element value of the matrix code is represented by 0 or 1; Step S302, constructing an objective function through the prediction model, and obtaining the fitness value of each individual in the initialized population through the objective function calculation; Step S303, selecting two individuals with the largest fitness values in the initialization population as parents; Step S304, generating a random number in the range of 0 to 1, and when the random number is less than a first threshold, performing a crossover operation that satisfies the constraint condition; The first threshold is a custom parameter; The crossover operation means randomly selecting the element value of one position in the matrix encoding of two individuals in the parent generation to exchange; Step S305, generating a random number in the range of 0 to 1, and when the random number is less than a second threshold, performing a mutation operation that satisfies the constraint condition; The first threshold is a custom parameter; The mutation operation means randomly selecting an individual in the parent generation to regenerate the element value of one position in the matrix encoding; Step S306, if it is determined that the current number of iterations is greater than or equal to the maximum number of iterations, the individual with the largest fitness value in the initialized population is used as the scheduling solution, otherwise, the process returns to step S302, where the maximum number of iterations is a custom parameter.
10. The enterprise digital intelligent decision-making system according to claim 9, characterized in that: The objective function is calculated as follows: Where Fit represents the fitness value of the individual, and They respectively represent the estimated cost and estimated completion time output by inputting the project data of the mth project to be assigned and the team data of the nth team into the prediction model, W1 and W2 represent the first weight parameter and the second weight parameter, respectively, both W1 and W2 are custom parameters and the sum of W1 and W2 is 1.
Citation Information
Patent Citations
Cost prediction method and device based on neural network model, medium and equipment
CN115423186A
Intelligent personnel allocation method based on grid management and storage medium
CN117787648A
Cited By
New energy station voltage regulation and control optimization system based on real-time monitoring
CN119921340A
