CN-121981490-A - Intelligent decision-making system based on artificial intelligence and big data

CN121981490ACN 121981490 ACN121981490 ACN 121981490ACN-121981490-A

Abstract

The invention relates to the technical field of intelligent decision and management, and particularly discloses an intelligent decision system based on artificial intelligence and big data, which comprises the following components: the method comprises the steps of acquiring real-time fusion data streams by a sensor of the Internet of things, storing the real-time fusion data streams into a time sequence database, defining and calculating a real-time energy efficiency index, constructing a dynamic digital twin body by combining the time sequence database, defining and calculating a system toughness index, judging whether scene switching is triggered by combining an identification operation scene, if so, carrying out simulation calculation on a dynamic threshold group by adopting the dynamic digital twin body, constructing an optimization problem model by taking the dynamic threshold group as a constraint target, carrying out multi-target optimization solving in the dynamic digital twin body by combining the real-time energy efficiency index, generating an executable optimization strategy package, solving the problem of serious disjoint between a static and isolated energy efficiency management system and an actual production operation environment of dynamic, variable and multi-target constraint, and realizing intelligent analysis management closed loop.

Inventors

YE YILONG
MO YIJING
ZHANG FEIFEI
XU HUI

Assignees

杭州智科飞创信息科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260210

Claims (9)

1. An intelligent decision-making system based on artificial intelligence and big data is characterized by comprising the following modules: The acquisition analysis module is used for deploying an Internet of things sensor to acquire real-time fusion data streams and store the real-time fusion data streams into the time sequence database, and defining and calculating a real-time energy efficiency index; a digital twin module for constructing dynamic digital twin body by combining the time sequence database; the dynamic constraint analysis module is used for defining and calculating a system toughness index, judging whether scene switching is triggered by combining with the identification of an operation scene, and carrying out simulation calculation on a dynamic threshold value group by adopting a dynamic digital twin body if the scene switching is triggered; And the optimization execution module is used for constructing an optimization problem model by taking the dynamic threshold value group as a constraint target, carrying out multi-target optimization solution in the dynamic digital twin body by combining the real-time energy efficiency index, and generating an executable optimization strategy package.
2. The intelligent decision-making system based on artificial intelligence and big data according to claim 1, wherein the real-time energy efficiency index is calculated by the following steps: And defining an analysis calculation period, wherein in the analysis calculation period, the real-time energy efficiency index is the ratio of the theoretical minimum energy consumption of effective production to the actual total energy consumption of the system, wherein the theoretical minimum energy consumption is determined by the product of the expected production of the current production planning work order and the theoretical minimum energy consumption of a product unit, and the actual total energy consumption of the system is obtained by querying a time sequence database.
3. The intelligent decision-making system based on artificial intelligence and big data according to claim 1, wherein the dynamic digital twin body is constructed by the following steps: The method comprises the steps of constructing three layers of model frames of equipment level, production line level and system level, wherein each layer of model instance comprises static attribute, dynamic state variable and interaction interface, performing association fusion on a computer aided design model and the model frames, establishing real-time mapping of the dynamic state variable and real-time fusion data flow through a data subscription channel, driving and updating by a state synchronization engine, and deploying and operating an equipment health assessment model in a dynamic digital twin body.
4. The intelligent decision system based on artificial intelligence and big data according to claim 3, wherein the device health evaluation model is constructed by the following steps: The method comprises the steps of adopting a framework formed by a parallelization multichannel one-dimensional convolutional neural network and a hierarchical long-short-term memory network, inputting a time sequence of equipment state data in a preset period, extracting features through a convolutional layer, capturing time sequence dependency relations by the long-short-term memory network, arranging a main regression head and an auxiliary classification head on an equipment health assessment model, respectively outputting normalized equipment health score and fault mode probability vectors, and training the equipment health assessment model based on acquired historical band marking data.
5. The intelligent decision system based on artificial intelligence and big data according to claim 1, wherein the system toughness index is calculated by the following method: Defining a toughness evaluation period, extracting a real-time energy efficiency index time sequence in the toughness evaluation period to form a performance standard curve, identifying performance attenuation events of which the real-time energy efficiency index is continuously lower than a preset performance base line threshold, and approximately calculating a toughness triangle area by the area surrounded by a start-stop moment and the formed performance standard curve for each performance attenuation event, wherein the system toughness index is calculated by subtracting the ratio of the sum of the toughness triangle areas of all events to a reference performance area, and the reference performance area is determined by multiplying the performance base line threshold by the length of the toughness evaluation period.
6. The intelligent decision system based on artificial intelligence and big data according to claim 1, wherein the judgment mode of whether to trigger scene switching is: Predefining four operation scenes of a steady-state high-efficiency scene, a planned disturbance scene, a sudden disturbance scene and an emergency supply scene, wherein each operation scene is associated with a preset constraint condition library; and continuously monitoring the real-time fusion data stream, the system toughness index and the external event signal through an operation scene recognition engine, judging the current operation scene according to a preset recognition rule, comparing the current scene with the previous moment, and triggering scene switching if the current scene is different from the previous moment.
7. The intelligent decision system based on artificial intelligence and big data according to claim 6, wherein the dynamic threshold group is calculated by: when scene switching is triggered, boundary conditions are set in the dynamic digital twin body according to constraint condition libraries corresponding to the identified operation scenes, and Monte Carlo simulation is operated; Based on the probability distribution result output by Monte Carlo simulation, calculating a dynamic threshold group comprises taking the low quantile of the real-time energy efficiency index result as a dynamic energy efficiency threshold, taking the high quantile of the total load power peak value as a dynamic load threshold, and taking the low quantile of the system toughness index result as a dynamic toughness threshold.
8. The intelligent decision system based on artificial intelligence and big data according to claim 7, wherein the construction mode of the optimization problem model is as follows: The dynamic digital twin body is taken as a simulation environment, decision variables comprise the set power of adjustable equipment, the energy storage charging and discharging power and time period, the start-stop state of non-critical load equipment and the flexible time offset of a non-highest priority work order, an objective function is the weighted summation of average real-time energy efficiency index, system toughness index and production plan delay penalty cost negative value in an optimization look-ahead period, the weight is determined based on a hierarchical analysis method according to an operation scene, and the constraint condition is integrated into a dynamic threshold group.
9. The intelligent decision system based on artificial intelligence and big data according to claim 8, wherein the generation mode of the optimization strategy package is as follows: An optimization solving engine integrating an improved non-dominant sorting genetic algorithm is adopted, candidate strategies are simulated in parallel in a dynamic digital twin body, objective function values are calculated, constraint is verified, a pareto optimal solution set is obtained through iteration, an optimal compromise strategy is selected according to decision rules, the optimal compromise strategy is decoded into a specific time-sequence executable instruction sequence, and the specific time-sequence executable instruction sequence is packaged into an optimization strategy package.

Description

Intelligent decision-making system based on artificial intelligence and big data Technical Field The invention relates to the technical field of intelligent decision and management, in particular to an intelligent decision system based on artificial intelligence and big data. Background Under the background of digital transformation and intelligent upgrading, enterprises have higher requirements on the real-time performance, the self-adaption performance and the global optimization capacity of operation management, a traditional decision system depends on historical data statistics, fixed rules and thresholds and a single-target optimization model, and the system is difficult to cope with complex reality scenes with dynamic variability, multiple constraints and multiple targets, and particularly when facing internal and external disturbance, the system often has the following limitations. Firstly, in the evaluation dimension, the traditional system lacks a quantitative evaluation mechanism for the dynamic toughness (namely the capability of resisting, absorbing and recovering disturbance) of the system, so that the attenuation and recovery process of the system performance cannot be reflected in real time when sudden faults, plan changes or external environment fluctuates, decision support is lagged, secondly, most of the systems adopt static threshold values and rule bases on the decision mechanism, the targets and constraints cannot be adaptively adjusted according to real-time operation scenes, so that the decision and actual business requirements are disjointed, the response is stiff, finally, on the optimization closed loop, the traditional method is often limited to local optimization or post analysis, the prospective multi-target collaborative optimization capability based on high-fidelity simulation is lacking, and executable global optimal strategies are difficult to generate on the premise of guaranteeing the toughness, the efficiency and the business constraints of the system. To overcome the above problems, there is a need for an intelligent system that can fuse real-time data, build dynamic models, quantify system toughness, and adaptively optimize decisions based on context. The invention provides an intelligent decision system based on artificial intelligence and big data, which realizes closed-loop intelligent decision from data to strategy by data acquisition of the Internet of things, dynamic digital twin construction, toughness index calculation, scene recognition and dynamic threshold generation and multi-objective optimization solution, and improves the overall operation efficiency and robustness of the system in a complex dynamic environment. Disclosure of Invention The invention aims to provide an intelligent decision system based on artificial intelligence and big data so as to solve the background problem. The invention can realize the aim by the following technical scheme that the intelligent decision system based on artificial intelligence and big data comprises: The acquisition analysis module is used for deploying an Internet of things sensor to acquire real-time fusion data streams and store the real-time fusion data streams into the time sequence database, and defining and calculating a real-time energy efficiency index; a digital twin module for constructing dynamic digital twin body by combining the time sequence database; the dynamic constraint analysis module is used for defining and calculating a system toughness index, judging whether scene switching is triggered by combining with the identification of an operation scene, and carrying out simulation calculation on a dynamic threshold value group by adopting a dynamic digital twin body if the scene switching is triggered; The optimization execution module is used for constructing an optimization problem model by taking the dynamic threshold value group as a constraint target, carrying out multi-target optimization solution in a dynamic digital twin body by combining the real-time energy efficiency index, and generating an executable optimization strategy package; further, the calculation mode of the real-time energy efficiency index is as follows: Defining an analysis calculation period, wherein in the analysis calculation period, the real-time energy efficiency index is the ratio of the theoretical minimum energy consumption of effective production to the actual total energy consumption of the system, wherein the theoretical minimum energy consumption is determined by the product of the expected production of the current production planning work order and the theoretical minimum energy consumption of a product unit, and the actual total energy consumption of the system is obtained by inquiring a time sequence database; further, the dynamic digital twin body is constructed in the following manner: Constructing three layers of model frames of equipment level, production line level and system level, wherein each layer of mo