China
Africa
Explore chapters and articles related to this topic
Methodologies of Feature Representations
Published in Awais Ahmad Khan, Emad Abouel Nasr, Abdulrahman Al-Ahmari, Syed Hammad Mian, Integrated Process & Fixture Planning, 2018
Awais Ahmad Khan, Emad Abouel Nasr, Abdulrahman Al-Ahmari, Syed Hammad Mian
In the rule-based methods, fixture knowledge is represented in the form of rules to govern the fixture design process. The rules are represented as If and Then logical statements. In expert systems, the knowledge is often formulized as a set of rules; therefore, they are also called rule-based systems. Nee et al. [52] presented a rule-based expert system for an automatic fixture design in which geometric and surface information were extracted to determine the type of operation and number of setups required using a solid modeler. Locating, supporting, and clamping planes and points were identified using rule-based and mathematical analysis. Perremans [53] introduced a fixture planning phase that included positioning, clamping, and supporting faces and used them as the input of the expert system. On the basis of these inputs, the expert system generated a physical fixture design. The main idea in the implementation of this approach was the description of modular elements by means of form features. Jeng and Gill [54] presented a fixture design problem in a hierarchical design structure. The automatic CAFD system was developed and used to automatically generate the fixture configuration for each setup orientation. The process plan was used as the input to the system. As a rule-based approach, this system was implemented for selecting the locating and clamping surfaces, and an algorithm-based search strategy was developed to automatically generate the fixture configuration for the construction of building of modular fixtures.
A Neuro-Symbolic Hybrid Intelligent Architecture with Applications
Published in Lakhmi Jain, Anna Maria Fanelli, Recent Advances in Artificial Neural Networks, 2000
The (optional) rule based system represents the initial domain theory extracted from domain experts in a rule-based format. The acquired rules are mapped into an initial connectionist architecture with uniform structure. The (optional) statistical module analyzes the available data sets and extracts certain correlations between different input parameters and also between input parameters and output decisions. The extracted statistical information is used to provide the mapped initial connectionist architecture with first and higher order input-input and input-output correlation rules. It is also used to provide supplementary rules to an initial rule-based system.
Computer-Integrated Manufacturing in the Food Industry
Published in Gauri S. Mittal, Computerized Control Systems in the Food Industry, 2018
The rule-based paradigm is the most popular paradigm for knowledge representation. A rule-based system typically consists of a database, a rule base, and an inference engine. The database contains the facts known to the system. The rule base contains all the relationships between the objects in the form of rules of the form IF{conditions}THEN{actions}
A fuzzy rule-based fog–cloud computing for solar panel disturbance investigation
Published in Cogent Engineering, 2019
Suryono Suryono, Ainie Khuriati, Teddy Mantoro
Output of fuzzy logic system computation is then used as input of the rule-based system developed to determine disturbance on solar panels. Rule-based system is one example of artificial intelligent with the principle of knowledge manipulation to gather more useful information. The rule-based system in this solar panel diagnosis can automatically run inferences that determines conditions swiftly and then stores it in the data base. Data history is kept for decision support system in optimizing output power of PV panels. Rule-based system employs rules set by people and automatically executed by the computer. This system is suitable for noncomplex regulation systems, with more accurate and speedy results (Gegov, Petrov, Sanders, & Vatchova, 2017).
Automating versus augmenting intelligence
Published in Journal of Enterprise Transformation, 2018
William B. Rouse, James C. Spohrer
The 1980s saw the growth of expert systems, led by Edward Feigenbaum (1980). These rule-based systems were built from “knowledge engineering” with subject matter experts. DARPA's Pilot's Associate's Program emerged to leverage expert systems technology (Banks & Lizza, 1991). Our basic research on intelligent interfaces (discussed below) was funded by a variety of agencies; this DARPA program provided the means to bring the pieces together.
Smart support system of material procurement for waste reduction based on big data and predictive analytics
Published in International Journal of Logistics Research and Applications, 2021
Tsai-Chi Kuo, Chien-Yun Peng, Chien-Jou Kuo
In this study, two types of models were constructed. The first type, (backpropagation) and (deep neural network), was related to the artificial neural network (ANN), and the second type , was related to the decision tree model. Figure 3 shows both these models. M1. This model was developed using an ANN. It was dependent on the number of neurons and hidden layers and the frequency of the training and output layers. In this model, a hidden layer was constructed. There are two main processes performed in the training model: forward pass/propagation and backpropagation. The backpropagation process involved the following steps. First, the error signal for each layer was evaluated. Second, the error signal was used to compute the error gradients. Third, the layer parameters were updated using the error gradients with an optimisation algorithm such as gradient descent.M2. This model was developed using a deep neural network based on the ANN algorithm. The methodology of M2 was similar to that of M1; however, the output of M1 was included in the input for M2.M3. This model was developed using a decision tree (DT). It can be considered as a type of rule-based system. In this research, the C4.5 algorithm of the DT was used. The steps are as follows: First, it creates a tree node whose value was the chosen attribute. Second, child links were created from this node, where each link represented a unique value for the chosen attribute. Third, the child link values were used to further subdivide the instances into subclasses. Finally, we specified the classification of new instances based on this decision path.