Reinforcement learning is an excellent approach which is used in artificial intelligence,automatic control, etc. However, ordinary reinforcement learning algorithm, such as Q-learning with lookup table cannot cope wit...Reinforcement learning is an excellent approach which is used in artificial intelligence,automatic control, etc. However, ordinary reinforcement learning algorithm, such as Q-learning with lookup table cannot cope with extremely complex and dynamic environment due to the huge state space. To reduce the state space, modular neural network Q-learning algorithm is proposed, which combines Q-learning algorithm with neural network and module method. Forward feedback neural network, Elman neural network and radius-basis neural network are separately employed to construct such algorithm. It is revealed that Elman neural network Q-learning algorithm has the best performance under the condition that the same neural network training method, i.e. gradient descent error back-propagation algorithm is applied.展开更多
In order to achieve the smart home system for remote management and wireless transmission function in the smart home networking system design, the article will combined with ARM technology and ZigBee wireless network ...In order to achieve the smart home system for remote management and wireless transmission function in the smart home networking system design, the article will combined with ARM technology and ZigBee wireless network technology based on samsung＇s $3C2440 chip and some related peripheral devices to form into home gateway of the system control platform, and the home control network using the internal ZigBee wireless network technologies. Practice has proved that the system is stable, is worth promoting.展开更多
Smart grid is a concept and not a computer system or some sort of hardware. Therefore, smart grids can have multiple shapes because there is no single unified concept of what constitutes a ＂Smart Grid＂. Smart grid i...Smart grid is a concept and not a computer system or some sort of hardware. Therefore, smart grids can have multiple shapes because there is no single unified concept of what constitutes a ＂Smart Grid＂. Smart grid is considered a core concept in sustainable electricity generation and consumption while being projected as a silver-bullet to electric power system challenges. South Africa＇s dwindling reserve margin, high CO_2 emissions and future energy mix are outstanding reasons for her smart grid deployment. Who receives the credit for smart grid neology is contentious and smart grid definition is diverse. However, there is a consensus on its attributes or benefits. The establishment of SASGI （South Africa Smart Grid Initiative） and Eskom＇s commencement of deploying a hybrid smart grid model are proofs of the country＇s commitment towards smart grid concept and developments. Having recognized the relevance of smart grid concept and its developments, South Africa is presented with the opportunity to ＂leap-frog＂ and enhance the relevant proven applicable smart grid solutions.展开更多
By introducing multi-agent technology to the online teaching system and establishing a personalized online teaching system model, the intelligent and personalized problems in the traditional online teaching system are...By introducing multi-agent technology to the online teaching system and establishing a personalized online teaching system model, the intelligent and personalized problems in the traditional online teaching system are solved; the needs of students for learning are fulfilled; the intelligent and interactive functions of the online teaching system are improved; a new interactive online teaching model is implemented. In this paper, the online system design based on multi-agent system technology is raised, and the modules and functions of the system are defined.展开更多
The study presents possibilities for reconstruction of electric power supply systems in Bulgarian Black Sea resorts and possibilities to use statistical methods in energy planning. The paper shows the use of classic s...The study presents possibilities for reconstruction of electric power supply systems in Bulgarian Black Sea resorts and possibilities to use statistical methods in energy planning. The paper shows the use of classic statistical methods in combination with advanced digital measurement systems in order to obtain the correlation dependencies, nature of energy consumption and opportunities for energy forecasting. The main purpose of the study is to obtain statistical dependencies of the nature of power consumption and correlations between electricity consumption and ambient temperature in order to improve the accuracy of energy planning. The analysis includes application of energy management systems for proper energy planning, improving economical efficiency and reducing power and energy losses.展开更多
The ITS is becoming more and more important in the economic development of China. But most of the ITS used in Chinese major cities need the human to perform the supervision task. As a result, it consumes too much huma...The ITS is becoming more and more important in the economic development of China. But most of the ITS used in Chinese major cities need the human to perform the supervision task. As a result, it consumes too much human resources, and also can not achieve the satisfied supervision performance. Thus, in this paper, we will propose an automatic inspection system based on the Gaussian mixture statistics model to alleviate this kind of problem. The proposed method will utilize a Gaussian Mixture model to model the background, and then use the EM algorithm to update the model＇s coefficients frame by frame to make the model adapt to the changing environment. After successful modeling, we can extract out the foreground blocks from background blocks, and finally trigger the automatic alarming system by calculating the number of foreground blocks. From the experiment results, our proposed method can achieve considerable good results.展开更多
文摘Reinforcement learning is an excellent approach which is used in artificial intelligence,automatic control, etc. However, ordinary reinforcement learning algorithm, such as Q-learning with lookup table cannot cope with extremely complex and dynamic environment due to the huge state space. To reduce the state space, modular neural network Q-learning algorithm is proposed, which combines Q-learning algorithm with neural network and module method. Forward feedback neural network, Elman neural network and radius-basis neural network are separately employed to construct such algorithm. It is revealed that Elman neural network Q-learning algorithm has the best performance under the condition that the same neural network training method, i.e. gradient descent error back-propagation algorithm is applied.
文摘In order to achieve the smart home system for remote management and wireless transmission function in the smart home networking system design, the article will combined with ARM technology and ZigBee wireless network technology based on samsung＇s $3C2440 chip and some related peripheral devices to form into home gateway of the system control platform, and the home control network using the internal ZigBee wireless network technologies. Practice has proved that the system is stable, is worth promoting.
文摘Smart grid is a concept and not a computer system or some sort of hardware. Therefore, smart grids can have multiple shapes because there is no single unified concept of what constitutes a ＂Smart Grid＂. Smart grid is considered a core concept in sustainable electricity generation and consumption while being projected as a silver-bullet to electric power system challenges. South Africa＇s dwindling reserve margin, high CO_2 emissions and future energy mix are outstanding reasons for her smart grid deployment. Who receives the credit for smart grid neology is contentious and smart grid definition is diverse. However, there is a consensus on its attributes or benefits. The establishment of SASGI （South Africa Smart Grid Initiative） and Eskom＇s commencement of deploying a hybrid smart grid model are proofs of the country＇s commitment towards smart grid concept and developments. Having recognized the relevance of smart grid concept and its developments, South Africa is presented with the opportunity to ＂leap-frog＂ and enhance the relevant proven applicable smart grid solutions.
文摘By introducing multi-agent technology to the online teaching system and establishing a personalized online teaching system model, the intelligent and personalized problems in the traditional online teaching system are solved; the needs of students for learning are fulfilled; the intelligent and interactive functions of the online teaching system are improved; a new interactive online teaching model is implemented. In this paper, the online system design based on multi-agent system technology is raised, and the modules and functions of the system are defined.
文摘The study presents possibilities for reconstruction of electric power supply systems in Bulgarian Black Sea resorts and possibilities to use statistical methods in energy planning. The paper shows the use of classic statistical methods in combination with advanced digital measurement systems in order to obtain the correlation dependencies, nature of energy consumption and opportunities for energy forecasting. The main purpose of the study is to obtain statistical dependencies of the nature of power consumption and correlations between electricity consumption and ambient temperature in order to improve the accuracy of energy planning. The analysis includes application of energy management systems for proper energy planning, improving economical efficiency and reducing power and energy losses.
文摘The ITS is becoming more and more important in the economic development of China. But most of the ITS used in Chinese major cities need the human to perform the supervision task. As a result, it consumes too much human resources, and also can not achieve the satisfied supervision performance. Thus, in this paper, we will propose an automatic inspection system based on the Gaussian mixture statistics model to alleviate this kind of problem. The proposed method will utilize a Gaussian Mixture model to model the background, and then use the EM algorithm to update the model＇s coefficients frame by frame to make the model adapt to the changing environment. After successful modeling, we can extract out the foreground blocks from background blocks, and finally trigger the automatic alarming system by calculating the number of foreground blocks. From the experiment results, our proposed method can achieve considerable good results.
