In the era of big data,data-driven technologies are increasingly leveraged by industry to facilitate autonomous learning and intelligent decision-making.However,the challenge of“small samples in big data”emerges whe...In the era of big data,data-driven technologies are increasingly leveraged by industry to facilitate autonomous learning and intelligent decision-making.However,the challenge of“small samples in big data”emerges when datasets lack the comprehensive information necessary for addressing complex scenarios,which hampers adaptability.Thus,enhancing data completeness is essential.Knowledge-guided virtual sample generation transforms domain knowledge into extensive virtual datasets,thereby reducing dependence on limited real samples and enabling zero-sample fault diagnosis.This study used building air conditioning systems as a case study.We innovatively used the large language model(LLM)to acquire domain knowledge for sample generation,significantly lowering knowledge acquisition costs and establishing a generalized framework for knowledge acquisition in engineering applications.This acquired knowledge guided the design of diffusion boundaries in mega-trend diffusion(MTD),while the Monte Carlo method was used to sample within the diffusion function to create information-rich virtual samples.Additionally,a noise-adding technique was introduced to enhance the information entropy of these samples,thereby improving the robustness of neural networks trained with them.Experimental results showed that training the diagnostic model exclusively with virtual samples achieved an accuracy of 72.80%,significantly surpassing traditional small-sample supervised learning in terms of generalization.This underscores the quality and completeness of the generated virtual samples.展开更多
The importance and necessity of energy saving in the world have been discussed for many years,but achieving a logical and transparent solution is still one of the main challenges and problems of the world’s eco...The importance and necessity of energy saving in the world have been discussed for many years,but achieving a logical and transparent solution is still one of the main challenges and problems of the world’s economy.The rapid growth of energy consumption in the last two decades has caused the security of the domestic energy supply of buildings to face serious problems.In this research,first by entering parameters such as the type of materials,doors and windows,and the type of soil on the floor connected to the ground,etc.in the heat and cold load calculation software(HAP Carrier)as the design calculations and then in the second step entering the specifications inferred from the Iran’s national building code as a reference for energy saving calculations,calculations are performed and compared as the first criterion,and finally these two outputs are compared.The actual energy consumption and determination of the building energy consumption index are determined as another criterion,as well as the degree of deviation from the actual consumption.The results showed that the theoretical method and the thermal and refrigeration load calculations of the Zanjan Gas Company building have 6%difference in cooling load but the heating load is about 34%different,which means for cooling loads,the theoretical model can be used with high accuracy but for heating loads,the national building code needs fundamental changes.展开更多
Digital twin is regarded as the next-generation technology for the effective operation of heating,ventilation and air conditioning(HVAC)systems.It is essential to calibrate the digital twin models to match them closel...Digital twin is regarded as the next-generation technology for the effective operation of heating,ventilation and air conditioning(HVAC)systems.It is essential to calibrate the digital twin models to match them closely with real physical systems.Conventional real-time calibration methods cannot satisfy such requirements since the computation loads are beyond acceptable tolerances.To address this challenge,this study proposes a clustering compression-based method to enhance the computation efficiency of digital twin model calibration for HVAC systems.This method utilizes clustering algorithms to remove redundant data for achieving data compression.Moreover,a hierarchical multi-stage heuristic model calibration strategy is developed to accelerate the calibration of similar component models.Its basic idea is that once a component model is calibrated by heuristic methods,its optimal solution is utilized to narrow the ranges of parameter probability distributions of similar components.By doing so,the calibration process can be guided,so that fewer iterations would be used.The performance of the proposed method is evaluated using the operational data from an HVAC system in an industrial building.Results show that the proposed clustering compression-based method can reduce computation loads by 97%,compared to the conventional calibration method.And the proposed hierarchical heuristic model calibration strategy is capable of accelerating the calibration process after clustering and saves 14.6%of the time costs.展开更多
基金supported by the National Natural Science Foundation of China(No.62306281)the Natural Science Foundation of Zhejiang Province(Nos.LQ23E060006 and LTGG24E050005)the Key Research Plan of Jiaxing City(No.2024BZ20016).
文摘In the era of big data,data-driven technologies are increasingly leveraged by industry to facilitate autonomous learning and intelligent decision-making.However,the challenge of“small samples in big data”emerges when datasets lack the comprehensive information necessary for addressing complex scenarios,which hampers adaptability.Thus,enhancing data completeness is essential.Knowledge-guided virtual sample generation transforms domain knowledge into extensive virtual datasets,thereby reducing dependence on limited real samples and enabling zero-sample fault diagnosis.This study used building air conditioning systems as a case study.We innovatively used the large language model(LLM)to acquire domain knowledge for sample generation,significantly lowering knowledge acquisition costs and establishing a generalized framework for knowledge acquisition in engineering applications.This acquired knowledge guided the design of diffusion boundaries in mega-trend diffusion(MTD),while the Monte Carlo method was used to sample within the diffusion function to create information-rich virtual samples.Additionally,a noise-adding technique was introduced to enhance the information entropy of these samples,thereby improving the robustness of neural networks trained with them.Experimental results showed that training the diagnostic model exclusively with virtual samples achieved an accuracy of 72.80%,significantly surpassing traditional small-sample supervised learning in terms of generalization.This underscores the quality and completeness of the generated virtual samples.
文摘The importance and necessity of energy saving in the world have been discussed for many years,but achieving a logical and transparent solution is still one of the main challenges and problems of the world’s economy.The rapid growth of energy consumption in the last two decades has caused the security of the domestic energy supply of buildings to face serious problems.In this research,first by entering parameters such as the type of materials,doors and windows,and the type of soil on the floor connected to the ground,etc.in the heat and cold load calculation software(HAP Carrier)as the design calculations and then in the second step entering the specifications inferred from the Iran’s national building code as a reference for energy saving calculations,calculations are performed and compared as the first criterion,and finally these two outputs are compared.The actual energy consumption and determination of the building energy consumption index are determined as another criterion,as well as the degree of deviation from the actual consumption.The results showed that the theoretical method and the thermal and refrigeration load calculations of the Zanjan Gas Company building have 6%difference in cooling load but the heating load is about 34%different,which means for cooling loads,the theoretical model can be used with high accuracy but for heating loads,the national building code needs fundamental changes.
基金support of the National Natural Science Foundation of China (No.51978601 and No.52161135202).
文摘Digital twin is regarded as the next-generation technology for the effective operation of heating,ventilation and air conditioning(HVAC)systems.It is essential to calibrate the digital twin models to match them closely with real physical systems.Conventional real-time calibration methods cannot satisfy such requirements since the computation loads are beyond acceptable tolerances.To address this challenge,this study proposes a clustering compression-based method to enhance the computation efficiency of digital twin model calibration for HVAC systems.This method utilizes clustering algorithms to remove redundant data for achieving data compression.Moreover,a hierarchical multi-stage heuristic model calibration strategy is developed to accelerate the calibration of similar component models.Its basic idea is that once a component model is calibrated by heuristic methods,its optimal solution is utilized to narrow the ranges of parameter probability distributions of similar components.By doing so,the calibration process can be guided,so that fewer iterations would be used.The performance of the proposed method is evaluated using the operational data from an HVAC system in an industrial building.Results show that the proposed clustering compression-based method can reduce computation loads by 97%,compared to the conventional calibration method.And the proposed hierarchical heuristic model calibration strategy is capable of accelerating the calibration process after clustering and saves 14.6%of the time costs.
