Dynamic scheduling training decision support system assisted by artificial intelligence
Through data collection, preprocessing, decision analysis and real-time scheduling adjustment, the problem of strong dependence on input data by existing systems is solved, and efficient and flexible dynamic scheduling decision support is achieved.
Patent Information
- Application Number
- CN202510504900.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-22
- Publication Date
- 2025-08-15
Smart Images
Figure CN120494338A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of dynamic scheduling training decision support, and in particular to an artificial intelligence-assisted dynamic scheduling training decision support system. Background Art
[0002] The "AI-assisted dynamic scheduling and training decision support system" is a system based on artificial intelligence technology that aims to optimize scheduling and training decisions. Its basic structure includes data input, data processing, decision support, and feedback optimization modules. First, the data input module is responsible for collecting data such as student information, course schedules, and resource availability, and passing it to the system. Next, the data processing module uses machine learning and data mining algorithms to analyze the input data and unearth potential rules and patterns. The decision support module generates the optimal scheduling and training plan based on factors such as real-time demand and resource status. Finally, the feedback and optimization module continuously adjusts and optimizes the system algorithm by monitoring the execution effect of decisions in real time to improve the accuracy and responsiveness of future decisions. The principle of this system relies on learning from historical data and using AI models for dynamic adjustment to ensure that the best decision is made in a changing environment.
[0003] While the system boasts significant intelligent advantages in scheduling optimization and decision support, it still suffers from certain drawbacks in practical applications. The system's decision-making effectiveness is critically dependent on the quality and integrity of input data. Errors or incompleteness in input data can prevent the system from making accurate predictions and adjustments. Because dynamic scheduling involves multiple real-time variables, the system may be unable to fully respond to emergencies or complex situations, resulting in inflexible and inefficient responses. Summary of the Invention
[0004] In response to the shortcomings of the existing technology, the present invention provides an artificial intelligence-assisted dynamic scheduling training decision support system, which solves the problem that the system's decision-making effect is extremely dependent on the quality and integrity of the input data; since dynamic scheduling involves multiple real-time variables, the system cannot fully cope with emergencies or complex situations, resulting in insufficient flexibility and efficiency in response.
[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: an artificial intelligence-assisted dynamic scheduling training decision support system, comprising:
[0006] Data collection module, used to collect student information, resource status, task requirements, and external environment data in real time;
[0007] Data preprocessing module, used to clean, correct errors and fill missing values in the collected data;
[0008] The decision analysis module analyzes the processed data and generates decision solutions based on deep learning models, reinforcement learning models, and predictive analysis algorithms;
[0009] The model optimization module learns the errors in the decision-making process through a feedback mechanism and optimizes the model parameters and decision-making strategies;
[0010] Real-time scheduling adjustment module, dynamically adjusts scheduling and training decisions based on real-time data changes and preset emergency response mechanisms;
[0011] The result feedback and monitoring module continuously monitors the decision-making effects and provides feedback to the decision analysis module for model improvement and adjustment.
[0012] Preferably, the decision analysis module adopts a deep learning model based on a neural network to perform feature extraction and modeling on the input data to accurately predict future demand and resource allocation. The data preprocessing module adopts anomaly detection and missing value interpolation algorithms to improve data quality and ensure the accuracy and completeness of the input data.
[0013] Preferably, the decision analysis module further includes a reinforcement learning-based model for self-adjustment and optimization according to actual scheduling execution results to cope with real-time changing environments.
[0014] Preferably, the real-time scheduling adjustment module automatically adjusts resource scheduling and training plans according to a preset emergency response mechanism when an anomaly is detected. The real-time scheduling adjustment module includes a real-time monitoring module for capturing system status changes and making decision adjustments at critical moments to respond to emergencies.
[0015] Preferably, the model optimization module automatically optimizes model parameters by comparing historical data and real-time feedback data to improve decision accuracy.
[0016] Preferably, the decision analysis module adopts a reasoning method based on fuzzy logic to solve the decision-making problem of the system when facing uncertainty and fuzzy data. The data acquisition module further includes a sensor interface and an external data interface for acquiring data from other systems and devices in real time.
[0017] Preferably, the output of the deep learning model is the optimal solution for scheduling and training decisions, formula:
[0018] Y = f(WX + b);
[0019] in,
[0020] Y is the model output, i.e. the prediction result;
[0021] f is the activation function, such as ReLU or sigmoid;
[0022] W is the weight matrix, which indicates how well the model learns the input data;
[0023] X is the input feature vector, which represents the historical data of the system;
[0024] b is the bias term.
[0025] Preferably, the reinforcement learning model is used to self-adjust and optimize according to the actual scheduling execution results to adapt to the dynamically changing environment. Formula:
[0026]
[0027] in,
[0028] Q(s,a) is the value function of the current state s and action a;
[0029] α is the learning rate, which determines the degree of influence of new information on old information;
[0030] r is the immediate reward in the current state;
[0031] γ is the discount factor, which indicates the degree of discount of future rewards;
[0032] s ′ is the new state, a ′ For the next step.
[0033] The present invention provides an artificial intelligence-assisted dynamic scheduling training decision support system. It has the following beneficial effects:
[0034] This AI-assisted dynamic scheduling training decision support system significantly improves the quality and integrity of input data through the anomaly detection and missing value interpolation algorithms in the data preprocessing module, resolving the problem of decision-making errors caused by inaccurate or incomplete data in existing systems. Precise feature extraction and data cleaning ensure that the system is able to perform decision analysis based on high-quality data. Combining the feature extraction of deep learning models with the self-adjustment capabilities of reinforcement learning models, the system can make accurate scheduling decisions based on real-time changing environments and continuously optimize strategies through reinforcement learning to cope with complex real-time scheduling needs. Furthermore, the use of a fuzzy logic-based reasoning method enables the system to cope with uncertainty and fuzzy data, better adapt to dynamic changes and abnormal situations, and enhance the robustness and flexibility of the system.
[0035] The system's real-time scheduling adjustment module and model optimization module effectively enhance decision-making flexibility and efficiency. The real-time monitoring module can quickly capture changes in system status and automatically adjust resource scheduling and training plans at critical moments to effectively respond to emergencies or complex situations. The model optimization module automatically optimizes model parameters by comparing historical data with real-time feedback data, enabling the system to maintain high decision-making accuracy and responsiveness in a constantly changing environment. This design enables the system to demonstrate strong adaptability and efficient decision-making support in dynamic scheduling scenarios, providing accurate and efficient support in complex and dynamic scheduling environments. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] Figure 1 It is a structural schematic diagram of the present invention. DETAILED DESCRIPTION
[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0038] Example 1
[0039] like Figure 1 As shown, an embodiment of the present invention provides an artificial intelligence-assisted dynamic scheduling training decision support system, including a data acquisition module for real-time collection of trainee information, resource status, task requirements, and external environment data.
[0040] The data preprocessing module is used to clean, correct errors and fill missing values in the collected data.
[0041] The decision analysis module analyzes the processed data and generates decision plans based on deep learning models, reinforcement learning models and predictive analysis algorithms. The decision analysis module uses a deep learning model based on neural networks to extract features and model the input data to accurately predict future demand and resource allocation. The data preprocessing module uses anomaly detection and missing value interpolation algorithms to improve data quality and ensure the accuracy and completeness of input data. The decision analysis module further includes a reinforcement learning-based model for self-adjustment and optimization based on actual scheduling execution results to cope with real-time changing environments. The decision analysis module uses a fuzzy logic-based reasoning method to solve the system's decision-making problems when facing uncertainty and fuzzy data. The data acquisition module further includes a sensor interface and an external data interface for acquiring data from other systems and devices in real time.
[0042] The model optimization module learns the errors in the decision-making process through the feedback mechanism and optimizes the model parameters and decision-making strategies. The model optimization module automatically optimizes the model parameters by comparing historical data and real-time feedback data to improve the accuracy of decision-making.
[0043] The real-time scheduling adjustment module dynamically adjusts scheduling and training decisions based on real-time data changes and preset emergency response mechanisms. The real-time scheduling adjustment module automatically adjusts resource scheduling and training plans when an anomaly is detected based on the preset emergency response mechanism. The real-time scheduling adjustment module includes a real-time monitoring module, which is used to capture changes in system status and make decision adjustments at critical moments to respond to emergencies.
[0044] The result feedback and monitoring module continuously monitors the decision-making effect and provides feedback to the decision analysis module for model improvement and adjustment. The output of the deep learning model is the optimal solution for scheduling and training decisions. The formula is:
[0045] Y = f(WX + b);
[0046] in,
[0047] Y is the model output, i.e. the prediction result;
[0048] f is the activation function, such as ReLU or sigmoid;
[0049] W is the weight matrix, which indicates how well the model learns the input data;
[0050] X is the input feature vector, which represents the historical data of the system;
[0051] b is the bias term.
[0052] The reinforcement learning model is used to self-adjust and optimize according to the actual scheduling execution results to adapt to the dynamically changing environment. The formula is:
[0053]
[0054] in,
[0055] Q(s,a) is the value function of the current state s and action a;
[0056] α is the learning rate, which determines the degree of influence of new information on old information;
[0057] r is the immediate reward in the current state;
[0058] γ is the discount factor, which indicates the degree of discount of future rewards;
[0059] s ′ is the new state, a ′ For the next step.
[0060] Experimental Examples
[0061] This experiment aims to verify the effectiveness and superiority of the proposed AI-assisted dynamic scheduling training decision support system in terms of real-time data acquisition, data processing, decision optimization, and scheduling adjustments. This experiment will verify the system's decision-making accuracy, response speed, and flexibility, particularly in the face of dynamic changes and emergencies.
[0062] Experimental methods:
[0063] This experiment was conducted on a high-performance computer with 16GB of memory, an Intel i7 processor, and an NVIDIA GTX 2080 graphics card.
[0064] Software environment:
[0065] The operating system is Ubuntu 20.04, Python version 3.8 is used, the deep learning framework uses TensorFlow, and the reinforcement learning algorithm is based on OpenAI Gym.
[0066] Dataset:
[0067] Student information: includes the student’s basic personal information (such as age, occupation, learning progress) and historical training records.
[0068] Resource status: including the number of available instructors, equipment status, and availability of training venues.
[0069] Task requirements: including students’ specific needs for the course and the urgency of the course content.
[0070] External environmental data: including weather, traffic conditions, and public emergencies.
[0071] The data collected historical data from 500 trainees, 10 training venues and 50 devices, including training tasks and resource scheduling records within 5 months.
[0072] Experimental process: a. Data collection and preprocessing:
[0073] The system collects data through sensors and external interfaces, and performs anomaly detection, missing value interpolation, and standardization on the collected data through the data preprocessing module.
[0074] For example, if there is missing data in a student’s training plan, the system uses an interpolation algorithm to fill in the missing position with the average value of historical data.
[0075] b. Decision Analysis:
[0076] A deep learning model is used to model student needs and resource status, and scheduling decisions are optimized based on reinforcement learning.
[0077] Deep learning model output formula:
[0078] Deep learning model output formula:
[0079] Assume that the output calculated by the model is:
[0080] Y=0.85 (i.e., the decision quality score of the decision plan).
[0081] The reinforcement learning model uses the formula:
[0082] Assume that the value function output by the model is:
[0083] Q(s,a)=0.75 (i.e., the quality score of the current scheduling plan)
[0084] Scheduling and real-time adjustments:
[0085] The system schedules resources based on the scheduling plan generated by the decision analysis module.
[0086] For example, the original plan was to arrange 10 trainees to receive training under the guidance of one lecturer, but during the experiment, the system discovered through real-time monitoring that a trainee suddenly fell ill. The system automatically adjusted and quickly deployed a lecturer from other sessions, ensuring the smooth progress of the training plan.
[0087] d. Feedback and optimization:
[0088] After implementing a scheduling decision, the system evaluates the scheduling effect through the results feedback and monitoring module, using historical data and real-time feedback information to optimize the model. The system continuously adjusts the model's learning parameters to improve the accuracy of future scheduling decisions.
[0089] Experimental data:
[0090]
[0091] Task Need Rating: Indicates the urgency of the student's need for the course, with a rating range of 1 (low) to 5 (high).
[0092] Device Status Rating: The operating condition of the device, with a rating range of 1 (bad) to 5 (good).
[0093] Training venue status rating: The availability of the training venue, rated on a scale of 1 (poor) to 5 (good).
[0094] Prediction scheduling quality (deep learning): The quality score of the prediction scheduling of training tasks using the deep learning model.
[0095] Reinforcement learning decision value: The scheduling decision quality score obtained by the reinforcement learning model.
[0096] Experimental results:
[0097] Decision accuracy:
[0098] Comparative experiments have shown that the system's predictions, based on deep learning and reinforcement learning models, improve decision-making by 18% compared to traditional manual scheduling methods. For example, in the experiment, the system was able to make timely adjustments to equipment scheduling when faced with an unexpected equipment failure, ensuring that training tasks were not affected. Traditional methods, however, experience scheduling delays, which can lead to increased waiting times for trainees.
[0099] Response speed:
[0100] Experiments have shown that after an emergency occurs, the system's real-time scheduling and adjustment module can respond and adjust the scheduling plan within 5 seconds, successfully handling sudden incidents such as student illness and equipment failure. This response speed far exceeds the manual scheduling time of traditional systems.
[0101] flexibility:
[0102] The system can handle dynamically changing environments, such as responding to emergencies and real-time changes in equipment status, and can flexibly respond to uncertain environments. For example, when a sudden venue adjustment was needed, the system quickly made adjustments using a reinforcement learning model, ensuring the efficient use of teaching resources.
[0103] in conclusion:
[0104] The results of this experiment demonstrate that the AI-assisted dynamic scheduling training decision support system of the present invention can efficiently and flexibly perform resource scheduling and decision optimization in complex scheduling environments. This system not only improves decision accuracy and reduces the impact of unexpected events, but also, through real-time scheduling adjustments and feedback mechanisms, provides adaptive capabilities, enabling continuous optimization of decision-making strategies and improving the overall utilization efficiency of training resources.
[0105] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.
Claims
1. An artificial intelligence-assisted dynamic scheduling training decision support system, characterized in that: include: Data collection module, used to collect student information, resource status, task requirements, and external environment data in real time; Data preprocessing module, used to clean, correct errors and fill missing values in the collected data; The decision analysis module analyzes the processed data and generates decision solutions based on deep learning models, reinforcement learning models, and predictive analysis algorithms; The model optimization module learns the errors in the decision-making process through a feedback mechanism and optimizes the model parameters and decision-making strategies; Real-time scheduling adjustment module, dynamically adjusts scheduling and training decisions based on real-time data changes and preset emergency response mechanisms; The result feedback and monitoring module continuously monitors the decision-making effects and provides feedback to the decision analysis module for model improvement and adjustment.
2. The artificial intelligence-assisted dynamic scheduling training decision support system according to claim 1, characterized in that: The decision analysis module adopts a deep learning model based on a neural network, and the data preprocessing module adopts anomaly detection and missing value interpolation algorithms.
3. The artificial intelligence-assisted dynamic scheduling training decision support system according to claim 1, characterized in that: The decision analysis module further includes a reinforcement learning-based model for self-adjustment and optimization according to actual scheduling execution results.
4. The artificial intelligence-assisted dynamic scheduling training decision support system according to claim 1, characterized in that: The real-time scheduling adjustment module automatically adjusts resource scheduling and training plans according to a preset emergency response mechanism when an anomaly is detected. The real-time scheduling adjustment module includes a real-time monitoring module for capturing system status changes.
5. The artificial intelligence-assisted dynamic scheduling training decision support system according to claim 1, characterized in that: The model optimization module automatically optimizes model parameters by comparing historical data and real-time feedback data.
6. The artificial intelligence-assisted dynamic scheduling training decision support system according to claim 1, characterized in that: The decision analysis module adopts a reasoning method based on fuzzy logic, and the data acquisition module further includes a sensor interface and an external data interface.
7. The artificial intelligence-assisted dynamic scheduling training decision support system according to claim 2, characterized in that: The output of the deep learning model is the optimal solution for scheduling and training decisions, formula: Y = f(WX + b); in, Y is the model output, i.e. the prediction result; f is the activation function, such as ReLU or sigmoid; W is the weight matrix, which indicates how well the model learns the input data; X is the input feature vector, which represents the historical data of the system; b is the bias term.
8. The artificial intelligence-assisted dynamic scheduling training decision support system according to claim 1, characterized in that: The reinforcement learning model is used to self-adjust and optimize according to the actual scheduling execution results to adapt to the dynamically changing environment. The formula is: in, Q(s,a) is the value function of the current state s and action a; α is the learning rate, which determines the degree of influence of new information on old information; r is the immediate reward in the current state; γ is the discount factor, which indicates the degree of discount of future rewards; s ′ is the new state, a ′ For the next step.