CIM intelligent decision-making method and system based on multi-modal AI large model

Abstract

The invention belongs to the field of CIM intelligent decision making, particularly relates to a CIM intelligent decision making method and system based on a multi-modal AI large model, and aims to solve the problems that the implementation effect of decision making is inconvenient to evaluate, the effect is poor and optimization cannot be carried out in time in the prior art, the following scheme is provided: the method comprises the following steps: S1, building a CIM intelligent decision making system framework, the CIM intelligent decision-making system framework comprises a data acquisition module, a multi-mode AI large model, a data analysis module, a prediction module, a decision-making suggestion module, an execution feedback module, an effect evaluation module, a decision-making optimization module, an evaluation standard setting module and a storage module. And S2, collecting the multi-modal data influencing the CIM intelligent decision through a data collection module, collecting the feedback data in the implementation process, evaluating the decision implementation effect according to the feedback data, making a judgment according to the evaluation result, and adjusting and optimizing the decision scheme.

Description

Technical Field

[0001] The present invention relates to the technical field of CIM intelligent decision-making, and in particular to a CIM intelligent decision-making method and system based on a multimodal AI large model. Background Art

[0002] CIM intelligent decision-making is a method that uses advanced computer technology and artificial intelligence algorithms to provide comprehensive decision support to decision makers by analyzing and mining big data. CIM system composition: CIM consists of multiple systems such as MES, EAP, SPC, etc. It is the sum of a series of software tool products for automated information management in advanced manufacturing plants, with flexible architecture deployment and real-time data management capabilities. Application areas: CIM intelligent decision-making is not only used in the manufacturing industry, but also widely used in urban planning, construction, management and other fields. By integrating multi-source information, it provides comprehensive, real-time and accurate urban information to support more efficient decision-making. Decision support: CIM intelligent decision-making assistance can help companies predict market trends and optimize their businesses, thereby improving their ability to adapt to market competition and decision-making efficiency;

[0003] In the existing technology, after CIM intelligent decision-making makes a decision, it is not convenient to evaluate the implementation effect of the decision, and the effect cannot be optimized in time. Therefore, we propose a CIM intelligent decision-making method and system based on a multimodal AI large model to solve the above problems. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings in the prior art that it is inconvenient to evaluate the implementation effect of the decision and the effect is not good and cannot be optimized in time, and to propose a CIM intelligent decision-making method and system based on a multimodal AI large model.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A CIM intelligent decision-making method based on a multimodal AI large model includes the following steps:

[0007] S1. Build the CIM intelligent decision-making system framework, which includes data acquisition module, multimodal AI large model, data analysis module, prediction module, decision opinion module, execution feedback module, effect evaluation module, decision optimization module, evaluation standard setting module, and storage module;

[0008] S2. Collect multimodal data that affect CIM intelligent decision-making through the data acquisition module, process the multimodal data through the multimodal AI big model, and convert data of different modalities into a unified representation form;

[0009] S3. Analyze the processed data through the data analysis module, apply statistical analysis, machine learning and data mining technology to mine data features, and the prediction module conducts in-depth predictive analysis based on the analysis results to mine the laws and trends behind the data;

[0010] S4, the decision-making opinion module gives decision-making opinions based on the rules and trends behind the data, implements the decision-making opinions through the execution feedback module, and collects feedback data during the implementation process;

[0011] S5. The effect evaluation module evaluates the effect of decision implementation based on the feedback data, and the decision optimization module makes judgments based on the evaluation results and adjusts and optimizes the decision plan.

[0012] Preferably, the specific workflow of the effect evaluation module is as follows: Determine the evaluation criteria: Before the decision is implemented, it is necessary to clarify the goals of the decision and the expected effects. According to these goals and effects, set specific and quantifiable evaluation criteria so as to accurately measure the effect of the decision later; Data collection and collation: During and after the implementation of the decision, continuously collect relevant data, including data on the execution of the decision, data on the impact, and data on possible external factors, and collate these data to ensure the accuracy and completeness of the data for subsequent analysis and evaluation; Comparative analysis: Compare the actual collected data with the preset evaluation criteria to analyze whether the decision has achieved the expected effect. Through comparative analysis, the advantages and disadvantages of the decision, as well as the possible reasons, can be identified; Feedback and adjustment: Based on the evaluation results, provide timely feedback to decision makers to point out the effects of the decision and existing problems.

[0013] Preferably, the decision optimization module includes a decision optimization algorithm, which includes a biological evolution algorithm, an ant colony algorithm, and a genetic programming. Through the algorithm, the decision problem is modeled and analyzed to find the optimal decision that meets specific goals.

[0014] Preferably, the components of the decision optimization module are: decision variables: variables that can be controlled by decision makers; objective function: defines the goal that the decision maker wants to optimize, which is a mathematical expression of the decision variables; constraints: conditions that limit the value range or relationship of decision variables. According to the nature of the problem, the decision optimization model can be divided into linear programming, nonlinear programming, integer programming, and dynamic programming types; according to the risk preference of the decision maker, it can also be divided into deterministic models, stochastic models, and fuzzy models.

[0015] Preferably, the decision opinion module relies on data analysis and statistical principles, collects and organizes large amounts of data to reveal patterns and potential causal relationships, provides accurate basis for decision-making, uses logic and model reasoning to assist decision-making, helps understand complex situations and optimize choices, and uses decision tables and decision trees to express decisions in an intuitive way, ensuring that all relevant factors are considered to improve the comprehensiveness and accuracy of decisions; Reinforcement learning mechanism: In a dynamic environment, the decision opinion module uses reinforcement learning to continuously adjust strategies to maximize returns through the interaction between the intelligent agent and the environment, thereby achieving adaptive decision-making.

[0016] Preferably, the principle of the prediction module is as follows: first, relevant data needs to be collected from various sources, and the collected data is cleaned, converted and normalized for preprocessing to eliminate noise, fill missing values ​​and unify the data format; feature extraction and selection: useful features are extracted from the preprocessed data, which can reflect the laws and patterns behind the data. Through the feature selection method, the features that have the greatest impact on the prediction target are screened out to improve the accuracy and efficiency of the prediction model; model construction and training: according to the characteristics of the data and the prediction target, a machine learning algorithm is selected, including linear regression, decision tree, and neural network; the model is trained using a training data set, the prediction error is minimized by adjusting the model parameters, and the generalization ability of the model is verified; the trained model is used to predict the test data set or new data to obtain the prediction result.

[0017] The present invention also proposes a CIM intelligent decision-making system based on a multimodal AI big model, including a CIM intelligent decision-making system framework. The CIM intelligent decision-making system framework includes a data acquisition module, a multimodal AI big model, a data analysis module, a prediction module, a decision opinion module, an execution feedback module, an effect evaluation module, a decision optimization module, an evaluation standard setting module, and a storage module. The CIM intelligent decision-making system framework is the above-mentioned CIM intelligent decision-making system framework.

[0018] Preferably, the multimodal AI large model includes a data processing unit, a model architecture unit, an algorithm optimization unit and an output unit;

[0019] The data processing unit integrates data from different modalities so that they can be processed and analyzed under a unified framework;

[0020] The model architecture unit is used to achieve cross-modal understanding and generation capabilities, including the following parts: Multimodal input layer: responsible for receiving input data from different modalities; Shared representation layer: unify the data of different modalities and extract common features; Cross-modal output layer: output corresponding results according to task requirements;

[0021] The algorithm optimization unit is used to improve the parameter update method of the model, increase the generalization ability of the model, or reduce the training cost of the model.

[0022] Preferably, the evaluation standard setting module includes an identity authentication unit, a standard setting unit, a standard optimization unit and a transmission unit, the identity authentication unit is connected to the standard setting unit, the standard setting unit is connected to the standard optimization unit, and the standard optimization unit is connected to the transmission unit.

[0023] Preferably, the storage module is used to store decision opinions and decision optimization data, and the decision module is connected to a management module, which is used to manage the data stored in the storage module.

[0024] In the present invention, the CIM intelligent decision-making method and system based on a multimodal AI large model have the following beneficial effects:

[0025] The data acquisition module collects multimodal data that affect CIM intelligent decision-making, processes multimodal data through a multimodal AI big model, and converts data of different modalities into a unified representation. The data analysis module analyzes the processed data, uses statistical analysis, machine learning, and data mining techniques to mine data features, and the prediction module conducts in-depth predictive analysis based on the analysis results to mine the laws and trends behind the data.

[0026] The decision-making opinion module gives decision-making opinions based on the rules and trends behind the data, and executes the decision-making opinions through the execution feedback module, while collecting feedback data during the implementation process; the effect evaluation module evaluates the effect of decision implementation based on the feedback data, and the decision optimization module makes judgments based on the evaluation results and adjusts and optimizes the decision plan;

[0027] The present invention collects feedback data during the implementation process, evaluates the effect of decision implementation based on the feedback data, makes judgments based on the evaluation results, and adjusts and optimizes the decision plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A block diagram of a CIM intelligent decision-making system based on a multimodal AI large model proposed by the present invention;

[0029] Figure 2 A block diagram of a multimodal AI big model of a CIM intelligent decision-making system based on a multimodal AI big model proposed by the present invention;

[0030] Figure 3 A block diagram of an evaluation standard setting module of a CIM intelligent decision-making system based on a multimodal AI large model proposed by the present invention;

[0031] Figure 4A block diagram of an early warning module of Embodiment 3 of a CIM intelligent decision-making system based on a multimodal AI large model proposed by the present invention;

[0032] Figure 5 This is a block diagram of a state monitoring module of Embodiment 2 of a CIM intelligent decision-making system based on a multimodal AI large model proposed in the present invention. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0034] Embodiment 1

[0035] Reference Figure 1-Figure 3 , a CIM intelligent decision-making method based on a multimodal AI big model, comprising the following steps:

[0036] S1. Build the CIM intelligent decision-making system framework, which includes data acquisition module, multimodal AI large model, data analysis module, prediction module, decision opinion module, execution feedback module, effect evaluation module, decision optimization module, evaluation standard setting module, and storage module;

[0037] S2. Collect multimodal data that affect CIM intelligent decision-making through the data acquisition module, process the multimodal data through the multimodal AI big model, and convert data of different modalities into a unified representation form;

[0038] S3. Analyze the processed data through the data analysis module, apply statistical analysis, machine learning and data mining techniques to mine data features, and the prediction module conducts in-depth predictive analysis based on the analysis results to mine the laws and trends behind the data;

[0039] S4, the decision-making opinion module gives decision-making opinions based on the rules and trends behind the data, implements the decision-making opinions through the execution feedback module, and collects feedback data during the implementation process;

[0040] S5. The effect evaluation module evaluates the effect of decision implementation based on the feedback data, and the decision optimization module makes judgments based on the evaluation results and adjusts and optimizes the decision plan.

[0041] In this embodiment, the specific work flow of the effect evaluation module is as follows: Determine the evaluation criteria: Before the decision is implemented, it is necessary to clarify the goals of the decision and the expected effects. According to these goals and effects, set specific and quantifiable evaluation criteria so as to accurately measure the effect of the decision later; Data collection and collation: During and after the implementation of the decision, continuously collect relevant data, including data on the execution of the decision, data on the impact, and possible external factors, and collate these data to ensure the accuracy and completeness of the data for subsequent analysis and evaluation; Comparative analysis: Compare the actual collected data with the preset evaluation criteria to analyze whether the decision has achieved the expected effect. Through comparative analysis, the advantages and disadvantages of the decision, as well as the possible reasons, can be identified; Feedback and adjustment: Based on the evaluation results, provide timely feedback to the decision maker to point out the effect of the decision and the problems that exist.

[0042] In this embodiment, the decision optimization module includes a decision optimization algorithm, which includes a biological evolution algorithm, an ant colony algorithm, and a genetic programming algorithm. Through the algorithm, the decision problem is modeled and analyzed to find the optimal decision that meets specific goals.

[0043] In this embodiment, the components of the decision optimization module are: decision variables: variables that can be controlled by decision makers; objective function: defines the goal that the decision maker wants to optimize, which is a mathematical expression of the decision variables; constraints: conditions that limit the value range or relationship of decision variables. According to the nature of the problem, the decision optimization model can be divided into linear programming, nonlinear programming, integer programming, and dynamic programming types; according to the decision maker's risk preference, it can also be divided into deterministic models, stochastic models, and fuzzy models.

[0044] In this embodiment, the decision opinion module relies on data analysis and statistical principles, collects and organizes large amounts of data to reveal patterns and potential causal relationships, provides accurate basis for decision-making, uses logic and model reasoning to assist decision-making, helps understand complex situations and optimize choices, and uses decision tables and decision trees to express decisions in an intuitive way, ensuring that all relevant factors are considered to improve the comprehensiveness and accuracy of decisions; Reinforcement learning mechanism: In a dynamic environment, the decision opinion module uses reinforcement learning to continuously adjust strategies to maximize returns through the interaction between the intelligent agent and the environment, thereby achieving adaptive decision-making.

[0045] In this embodiment, the principle of the prediction module is as follows: first, relevant data needs to be collected from various sources, and the collected data is cleaned, converted and normalized for preprocessing to eliminate noise, fill missing values ​​and unify the data format; feature extraction and selection: useful features are extracted from the preprocessed data, which can reflect the laws and patterns behind the data. Through the feature selection method, the features that have the greatest impact on the prediction target are screened out to improve the accuracy and efficiency of the prediction model; model construction and training: according to the characteristics of the data and the prediction target, a machine learning algorithm is selected, including linear regression, decision tree, and neural network; the model is trained using a training data set, the prediction error is minimized by adjusting the model parameters, and the generalization ability of the model is verified; the trained model is used to predict the test data set or new data to obtain the prediction result.

[0046] This embodiment also proposes a CIM intelligent decision-making system based on a multimodal AI big model, including a CIM intelligent decision-making system framework. The CIM intelligent decision-making system framework includes a data acquisition module, a multimodal AI big model, a data analysis module, a prediction module, a decision opinion module, an execution feedback module, an effect evaluation module, a decision optimization module, an evaluation standard setting module, and a storage module.

[0047] In this embodiment, the multimodal AI large model includes a data processing unit, a model architecture unit, an algorithm optimization unit, and an output unit;

[0048] The data processing unit integrates data from different modalities so that they can be processed and analyzed under a unified framework;

[0049] The model architecture unit is used to achieve cross-modal understanding and generation capabilities, including the following parts: Multimodal input layer: responsible for receiving input data from different modalities; Shared representation layer: unify the data of different modalities and extract common features; Cross-modal output layer: output corresponding results according to task requirements;

[0050] The algorithm optimization unit is used to improve the parameter update method of the model, increase the generalization ability of the model, or reduce the training cost of the model.

[0051] In this embodiment, the evaluation standard setting module includes an identity authentication unit, a standard setting unit, a standard optimization unit and a transmission unit. The identity authentication unit is connected to the standard setting unit, the standard setting unit is connected to the standard optimization unit, and the standard optimization unit is connected to the transmission unit.

[0052] In this embodiment, the storage module is used to store decision opinions and decision optimization data. The decision module is connected to the management module, and the management module is used to manage the data stored in the storage module.

[0053] Embodiment 2

[0054] Reference Figure 1 , Figure 5 The difference between this embodiment and the first embodiment is that:

[0055] S1. Build the CIM intelligent decision-making system framework, which includes data acquisition module, multimodal AI big model, data analysis module, prediction module, decision opinion module, execution feedback module, effect evaluation module, decision optimization module, evaluation standard setting module, storage module, and status monitoring module;

[0056] S2. Collect multimodal data that affect CIM intelligent decision-making through the data acquisition module, process the multimodal data through the multimodal AI big model, and convert data of different modalities into a unified representation form;

[0057] S3. Analyze the processed data through the data analysis module, apply statistical analysis, machine learning and data mining techniques to mine data features, and the prediction module conducts in-depth predictive analysis based on the analysis results to mine the laws and trends behind the data;

[0058] S4, the decision-making opinion module gives decision-making opinions based on the rules and trends behind the data, implements the decision-making opinions through the execution feedback module, and collects feedback data during the implementation process;

[0059] S5. The effect evaluation module evaluates the effect of decision implementation based on the feedback data, and the decision optimization module makes judgments based on the evaluation results and adjusts and optimizes the decision plan;

[0060] S6. Monitor the working status of the multimodal AI large model through the status monitoring module. The status monitoring module includes a benchmark test unit, a diversity test unit, a robustness test unit, an efficiency test unit and a practical application test unit. Benchmark and diversity testing: evaluate the model performance through standard data sets and tasks, test the performance of the model on different types of data and tasks, and ensure the model's ability to handle various language phenomena and contexts; robustness testing: check the performance of the model when facing input data disturbances, such as spelling errors, grammatical errors, etc., to ensure the error tolerance and stability of the model; efficiency testing: test the operating efficiency of the model under different computing resources and hardware environments, and evaluate the inference speed, memory usage and expansion capabilities; practical application testing: test the application effect of the model in real scenarios, collect user feedback and performance indicators, and evaluate practicality and user satisfaction.

[0061] Embodiment 3

[0062] Reference Figure 1 , Figure 4 , Figure 5, the difference between this embodiment and the second embodiment is: S1, building a CIM intelligent decision-making system framework, the CIM intelligent decision-making system framework includes a data acquisition module, a multimodal AI large model, a data analysis module, a prediction module, a decision opinion module, an execution feedback module, an effect evaluation module, a decision optimization module, an evaluation standard setting module, a storage module, a status monitoring module and an early warning module;

[0063] S2. Collect multimodal data that affect CIM intelligent decision-making through the data acquisition module, process the multimodal data through the multimodal AI big model, and convert data of different modalities into a unified representation form;

[0064] S3. Analyze the processed data through the data analysis module, apply statistical analysis, machine learning and data mining techniques to mine data features, and the prediction module conducts in-depth predictive analysis based on the analysis results to mine the laws and trends behind the data;

[0065] S4, the decision-making opinion module gives decision-making opinions based on the rules and trends behind the data, implements the decision-making opinions through the execution feedback module, and collects feedback data during the implementation process;

[0066] S5. The effect evaluation module evaluates the effect of decision implementation based on the feedback data, and the decision optimization module makes judgments based on the evaluation results and adjusts and optimizes the decision plan;

[0067] S6. Monitor the working status of the multimodal AI large model through the status monitoring module. The status monitoring module includes a benchmark test unit, a diversity test unit, a robustness test unit, an efficiency test unit, and a practical application test unit. Benchmark and diversity testing: evaluate the model performance through standard data sets and tasks, test the performance of the model on different types of data and tasks, and ensure the model's ability to handle various language phenomena and contexts; robustness testing: check the performance of the model when facing input data disturbances, such as spelling errors, grammatical errors, etc., to ensure the error tolerance and stability of the model; efficiency testing: test the operating efficiency of the model under different computing resources and hardware environments, and evaluate the inference speed, memory usage and expansion capabilities; practical application testing: test the application effect of the model in real scenarios, collect user feedback and performance indicators, and evaluate practicality and user satisfaction;

[0068] S7. The monitoring data of the status monitoring module is extracted and analyzed through the early warning module. If the data does not meet the requirements, an early warning is issued. The early warning module includes a data extraction unit, a comparison and analysis unit, an early warning unit and a recording unit.

[0069] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A CIM intelligent decision-making method based on a multimodal AI large model, characterized in that: The following steps are involved: S1. Build the CIM intelligent decision-making system framework, which includes data acquisition module, multimodal AI large model, data analysis module, prediction module, decision opinion module, execution feedback module, effect evaluation module, decision optimization module, evaluation standard setting module, and storage module; S2. Collect multimodal data that affect CIM intelligent decision-making through the data acquisition module, process the multimodal data through the multimodal AI big model, and convert data of different modalities into a unified representation form; S3. Analyze the processed data through the data analysis module, apply statistical analysis, machine learning and data mining technology to mine data features, and the prediction module conducts in-depth predictive analysis based on the analysis results to mine the laws and trends behind the data; S4, the decision-making opinion module gives decision-making opinions based on the rules and trends behind the data, implements the decision-making opinions through the execution feedback module, and collects feedback data during the implementation process; S5. The effect evaluation module evaluates the effect of decision implementation based on the feedback data, and the decision optimization module makes judgments based on the evaluation results and adjusts and optimizes the decision plan.

2. According to claim 1, a CIM intelligent decision-making method based on a multimodal AI large model is characterized in that: The specific workflow of the effect evaluation module is as follows: Determine the evaluation criteria: Before the decision is implemented, it is necessary to clarify the goals of the decision and the expected effects. According to these goals and effects, set specific and quantifiable evaluation criteria so as to accurately measure the effect of the decision later; Data collection and collation: During and after the implementation of the decision, continuously collect relevant data, including data on decision execution, impact data, and possible external factor data, and organize these data to ensure the accuracy and completeness of the data for subsequent analysis and evaluation; Comparative analysis: Compare the actual collected data with the preset evaluation criteria to analyze whether the decision has achieved the expected effect. Through comparative analysis, the advantages and disadvantages of the decision, as well as the possible reasons, can be identified; Feedback and adjustment: Based on the evaluation results, provide timely feedback to decision makers, pointing out the effect of the decision and any problems that exist.

3. According to claim 2, a CIM intelligent decision-making method based on a multimodal AI large model is characterized in that: The decision optimization module includes a decision optimization algorithm, which includes a biological evolution algorithm, an ant colony algorithm, and a genetic programming algorithm. Through the algorithm, the decision problem is modeled and analyzed to find the optimal decision that meets a specific goal.

4. According to claim 3, a CIM intelligent decision-making method based on a multimodal AI large model is characterized in that: The components of the decision optimization module are: decision variables: representing the variables that the decision maker can control; objective function: defining the goal that the decision maker wants to optimize, which is the mathematical expression of the decision variable; Constraints: conditions that limit the range or relationship of decision variables. According to the nature of the problem, decision optimization models can be divided into linear programming, nonlinear programming, integer programming, and dynamic programming types; according to the decision maker's risk preference, they can also be divided into deterministic models, stochastic models, and fuzzy models.

5. According to claim 4, a CIM intelligent decision-making method based on a multimodal AI large model is characterized in that: The decision-making opinion module relies on data analysis and statistical principles. It collects and organizes large amounts of data to reveal patterns and potential causal relationships, provide accurate basis for decision-making, use logic and model reasoning to assist decision-making, help understand complex situations and optimize choices, and use decision tables and decision trees to express decisions in an intuitive way, ensuring that all relevant factors are considered to improve the comprehensiveness and accuracy of decisions; Reinforcement learning mechanism: In a dynamic environment, the decision-making opinion module adopts reinforcement learning, through the interaction between the intelligent agent and the environment, constantly adjusts the strategy to maximize the return, and realizes adaptive decision-making.

6. According to claim 5, a CIM intelligent decision-making method based on a multimodal AI large model is characterized in that: The principles of the prediction module are as follows: first, relevant data needs to be collected from various sources, and the collected data needs to be cleaned, converted and normalized for preprocessing to eliminate noise, fill in missing values ​​and unify the data format; feature extraction and selection: useful features are extracted from the preprocessed data, which can reflect the laws and patterns behind the data. Through feature selection methods, the features that have the greatest impact on the prediction target are screened out to improve the accuracy and efficiency of the prediction model; model construction and training: according to the characteristics of the data and the prediction target, machine learning algorithms are selected, including linear regression, decision trees, and neural networks; the model is trained using a training data set, the prediction error is minimized by adjusting the model parameters, and the generalization ability of the model is verified; the trained model is used to predict the test data set or new data to obtain the prediction results.

7. A CIM intelligent decision-making system based on a multimodal AI large model, characterized in that: It includes a CIM intelligent decision-making system framework, which includes a data acquisition module, a multimodal AI large model, a data analysis module, a prediction module, a decision opinion module, an execution feedback module, an effect evaluation module, a decision optimization module, an evaluation standard setting module, and a storage module. The CIM intelligent decision-making system framework is the CIM intelligent decision-making system framework described in any one of claims 1 to 6.

8. The CIM intelligent decision-making system based on a multimodal AI big model according to claim 7 is characterized in that: The multimodal AI big model includes a data processing unit, a model architecture unit, an algorithm optimization unit and an output unit; The data processing unit integrates data from different modalities so that they can be processed and analyzed under a unified framework; The model architecture unit is used to achieve cross-modal understanding and generation capabilities, including the following parts: Multimodal input layer: responsible for receiving input data from different modalities; Shared representation layer: unify the data of different modalities and extract common features; Cross-modal output layer: outputs corresponding results according to task requirements; The algorithm optimization unit is used to improve the parameter update method of the model, increase the generalization ability of the model, or reduce the training cost of the model.

9. The CIM intelligent decision-making system based on a multimodal AI large model according to claim 8 is characterized in that: The evaluation standard setting module includes an identity authentication unit, a standard setting unit, a standard optimization unit and a transmission unit. The identity authentication unit is connected to the standard setting unit, the standard setting unit is connected to the standard optimization unit, and the standard optimization unit is connected to the transmission unit.

10. The CIM intelligent decision-making system based on a multimodal AI big model according to claim 9 is characterized in that: The storage module is used to store decision opinions and decision optimization data. The decision module is connected to a management module, and the management module is used to manage the data stored in the storage module.

Images