Research Of Feedback-simulated Cognitive Intelligence Mechanism And Calculating Model For Uncertain Information Cognition Object

The simulation of humanconsciousness and thinking process is a challenging science.Aiming at the significant differenceproblem between traditional cognition system with open-loop mode and information interaction processfor human cognitionwith the characteristics of repeated comparison and inference,and the engineering difficulty of online real-time measurement for the recognizability and accuracy of cognitive system,a novel trinitarianresearch method of humanoid cognitive intelligence with uncertain information cognition object modeling,generalized cognition error measurement,and feedback-simulated regulating mechanism is proposed in this paper.Based on rough set theory,the patternize structure model and operating mechanism of uncertain information cognition object is established.Based on genalized error and entropy theory,a multi-level and variable granularity genalized cognition error evaluative system with entropy function indexes is built.Based on feedback control theory and cognitive knowledge granularity calculating tool,a cognitive intelligence operating mechanism with feedback-simulated mode is explored to establish.This researchtried to establish cognitive intelligencemechanism and calculating model to imitate the repeated comparison and inferencecharacteristics of humanoid information interaction mode for a kind of cognitive intelligence system with uncertain information cognition object and cognition result,which expands the development of artificial intelligence.The main results are list as follows:(1)Research of the structure and operating mechanism of feedback-simulated cognitive intelligencesystem withuncertain information cognition object.The structure mode and operating mechanism of feedback-simulated cognitive intelligence system with three layers and three sections intercoupling are constructed to simulate human cognitive characteristic of repeated interaction mode from coase elaboration to fine comparison.The layerd and sectionalized building targets of feedback-simulated cognitive intelligence system are givenfirstly.Then,thefunctional requirement and operating mechanism are standardized,and the representative forms and acquisition methods of cognitive knowledge are defined.Therefore,the patternized construction method is supplied to establish the feedback-simulated cognitive intelligence system with uncertain information cognition object.(2)Research of cognitive intelligence decision information system model with uncertain information cognition object.Against the problem that there is no general knowledge acquisition method can adapt to all kinds of dataset,rough set-based cognitive intelligence decision information system with uncertain information cognition object multiple information fusion and supervised learning is constructed.The entropy measurement index of cognitive knowledge fusion sufficiency and distinguishablity are given based on feature reduction.Under the condition of uncertain information oflimited domain,the general model of cognitive intelligence decision information system is provided with cognitive knowledge optimal representation with uncertain information cognition object.(3)Research of generalized cognitive error evaluation systemwith uncertain cognition processeffect and cognitive result.In view ofthe general problem that online and real-time measurement of recognizability and accuracy in the cognitive system can't be satisfied with the posterior evaluation of cognitive result credibility,based on cognitive intelligence decision information system,the cognitive error information system with uncertain information cognition object and the uncertain cognitive process effect and cognitive result is constructed.Three kinds of generalized cognition error based on entropy evaluation index are defined,such as-entropy of uncertain cognition result error,-entropy sequence similarity of uncertain cognition result error and cognitive knowledge granularity error.The online and real-time evaluative system for credibility of uncertain cognition process effect and cognition result is established toprovide the quantitative criterion for the feedback-simulated cognitive intelligence adjustment mechanism.(4)Research of cognitive intelligence mechanism with uncertain information cognition object.To aim at the problem of significant difference between traditional cognition system with open-loop mode and human cognition information interaction mode with the characteristics of repeated comparison and inference,the feedback-simulated cognition intelligence adjustment mechanism is constructed based on online and real-time measurement of generalized cognitive errors.The feedback-simulated adjustment rules and calculation method for cognitive knowledge granularity are defined.The mode of the interlayer regulation and within-layer optimization with multi-level and variable granularity is created.The feedback-simulated cognition intelligence adjustment mechanism for uncertain information cognition objectwith variable knowledge granularity is realized to provide an intelligent cognition mechanism for imitating the information interaction process ofhuman cognition with repeated comparison and inference.(5)The machine cognition application and validation for offline handwritten Chinese character image with single target and complete state.Offline handwritten Chinese character image recognition is considered as one of difficulties in the field of pattern recognition with the characteristics of large number of quantities,categories,similar words and writing styles.Based on the feedback-simulated cognition intelligence method proposed in this paper,the feedback-simulated cognition intelligence system for offline handwritten Chinese character image with the intelligent adjustment mechanism of cognitive knowledge granularity is studied.Various Chinese character images are chosen from SCUT-IRAC HCCLIB handwritten Chinese character image library as the cognitive objects.The MATLAB simulation experiment is used to validate the feasibility and effectiveness of our method.The experimental results show thatour method has obvious advantages over traditional open-loop method.Therefore,a new practical application method for offline handwritten Chinese character imageis established to replace “human recognition” by “machine recognition”.(6)The machine cognition application and validation for industrial rotary kiln burning state with multi-targets and relatively complete state.Aiming at the problem of poor dataset caused by complex industrial environment,based on the feedback-simulated cognitive intelligence method proposed in this paper,the feedback-simulated cognition intelligence system of industrial rotary kiln burning state based on flame image is researched.The rotary kiln flame image samples and measurement dataare collected from No.2 rotary kiln of a cement plant supplied by State Key Laboratory of Synthetical Automation for Process Industries,Northeastern University.The MATLAB simulation experiment is used to validate the feasibility and effectiveness of our method.The experiment results show that the flame image feature knowledge can be mined from macro to micro to cognize the samples around the dicision surface repeatedly.Therefore,a new engineering application method for cement sintering process monitoring and the clinker quality-based close-loop control with “machine monitoring” instead of “human monitoring” are explored.(7)The machine cognition application and validation forhuman health state with multi-targets and incomplete state.Specific to the problems of heterogeneous body samples,multifarious and random body health features,based on the feedback-simulated cognitive intelligence method proposed in this paper,the feedback-simulated cognitive intelligence system based on variant cognition knowledge granularity for human health state is studied.The bioelectrical impedance signal samples of random peoples provided by Institute of Intelligent Machines,Chinese Academy of Sciences are collected.The MATLAB simulation experiment is used to validate the feasibility and effectiveness of our method.The experiment results show that the health knowledge is presented and the health indexes are hazily matched in each cognitive knowledge granularity layer.Therefore,a new practical application method for the objective and accurate evaluation of human health state evaluation to replace “artificial diagnosis” by “machine diagnosis” is explored.(8)The application and validation for intelligent cognition and tracking control of all-weather photovoltaic generation with single target and incomplete state.In view of the PV cell tracking control problem caused by the nonlinearity,nonstationarity,large intermittent and randomness of the weather environment,based on the feedback-simulated cognitive intelligence method proposed in this paper,the intelligent cognition and tracking control system of all-weather photovoltaic generation is studied and the method of calculating the solar position based on thevariant granularity and feedback-simulated cognition for panoramic image is given.Experiments were carried out to verify the feasibility and effectiveness of our method by using a variety of PV cell modules and their servo mechanisms such as flat type and condenser type.The experimental results show that the corresponding control strategy can be carried out based on the cognitive knowledge of environment and system state for wind,snow,daytime,night,rainy,cloudy and sunny,and the photovoltaic efficiency was improved based on the accurate calculation of the solar position in the sunny tracking control with our three-layer and three-stage intelligent cognition and tracking control system model.A kind of new method for intelligent controlling application is explored to achieve the efficient,energy-saving and secure operation for the photovoltaic generation system.