242 results about "Inference machine" patented technology

A motor control device capable of suitably inhibiting the pulsating torque of an alternating-current motor is provided. The motor control device comprises an axis error inference machine (30) which infers an axis error based on a current value of an inverter, detected by a current sensor (2); a Fourier rectification machine (316a) which extracts an axis error vector from the time-base changes of the axis error; an integral control machine (316b) additionally provided with a circular amplitude limiter, which is used to calculate a correction current vector for offsetting the pulsating torque, wherein a circle with a set amplitude limiting value serving as the radius is used as the base for limiting the movement of the correction current vector, and the integral control machine (316b) additionally provided with the circular amplitude limiter performs the circular amplitude limiting for limiting the movement of the correction current vector so as to enable the deflection angle of the correction current vector to be close to the deflection angle of the axis error vector.

An intelligent decision supporting system and a method for making an intelligent decision are provided. The intelligent decision supporting system includes a multi-dimensional classifier comprising a plurality of classifiers that define different semantic standards and are trained based on the different semantic standards, for classifying a text by the semantic standards and for outputting a plurality of attributes of the text and a confidence rate of each of the plurality of attributes, a question submitting module for receiving the output of the multi-dimensional classifier, for forming a question based on the plurality of attributes of the text and the confidence rate of each attribute, and for submitting the question to an inference machine, the inference machine for receiving the question submitted by the question submitting module, for inquiring of a domain ontology knowledge library based on the question, and for providing an answer for the question to an decision reply module, a domain ontology knowledge library module for storing a domain ontology knowledge library related to an application domain of the intelligent decision supporting system, wherein the domain ontology knowledge library records description of rules for deriving decisions corresponding to the semantic standards of the multi-dimensional classifier, and the decision reply module for providing the answer for the question provided by the inference machine to the user.

The invention discloses a system for quantitative management of software defects, which comprises a defect prediction and evaluation module, a process fault diagnosis module and a software defect mending cost estimating module. The defect prediction and evaluation module is used for evaluating and predicting the amount of software defects. The process fault diagnosis module is used for simulating the propagation process of software process faults, expressing fuzzy fault knowledge and providing a target-drive-oriented backward reasoning mechanism and software product fault diagnosis. The software defect mending cost estimating module is used for researching a defect mending cost predicting method so as to calculate mending cost and accurately estimate the defect mending cost. By the system, intelligent evaluation on software defects is realized, efficiency and accuracy in software process fault diagnosis are improved, software process is more controllable, and development process is more highly mature.

The invention provides a network fault diagnosis system. The system comprises: a simple network management protocol (SNMP) communication module, a database management module, a core diagnosis module, a knowledge management module and a user interaction module. An expert system is used to carry out intelligent fault monitoring. And a hierarchy correlation knowledge representation model which is suitable for a network knowledge characteristic is designed and is used to carry out knowledge representation. A high-efficient inference machine rule matching algorithm is designed according to a network fault characteristic. The system is especially suitable for a small/medium-size local area network. The network fault monitoring and management can be intelligentized or half-intelligentized. An efficiency of network management can be improved and costs can be effectively reduced. The knowledge representation method can be conveniently used in the other engineering fields and has wide applicability.

A Chinese semantic proofreading method based on ontology consistency verification and reasoning includes: using the ontology learning technology to extract semantic contents from non-structural natural Chinese languages, and converting the extracted semantic contents into structural ontology forms; establishing field ontology databases, and using corresponding ontology databases according to different fields; building models integrating the formed Chinese semantic proofreading key technology, in a form of plug-in, into a grammar verifying tool or using other implementation forms such as independent Chinese semantic verification software; in the grammar verifying tool, using the consistency reasoning and verification mechanism based on description logic and contained in the ontology reasoning language to sequentially perform consistency verification and reasoning on the extracted semantic contents and correct field ontology database input into a reasoning machine according to preset orders, and labeling error labels on the Chinese semantic contents, with inconsistent logic, in the reasoning result. By the method, word-level and grammar-level Chinese proofreading can be achieved, and Chinese semantic proofreading in special fields can also be achieved.

The invention relates to a knowledge-based fault prediction expert system for a complex milling machine tool, which comprises a man-machine interface module, a data acquisition and preprocessing module, a knowledge acquisition module based on data mining, a self-adaptive fault prediction module, a fault diagnosis expert system, an interpreter, an inference machine, a prediction method library, an integrated knowledge library, a model library and a diagnosis module; wherein the man-machine interface module is used for transmitting real-detection data and historical data to other modules; the data acquisition and preprocessing module is used for preprocessing the received data and inputting the preprocessed data to the knowledge acquisition module for data mining; the knowledge acquisition module is connected with the fault diagnosis expert system through the fault prediction module; the fault prediction module and the fault diagnosis expert system input fault prediction and diagnosis results into the inference machine, and the inference machine search in the integrated knowledge library and the model library; and the fault prediction module establishes a selection target function, and the inference machine selects a proper prediction method for fault prediction from the prediction method library. The invention can be widely applied in the fault diagnosis of various numerically-controlled machine tools.

The invention discloses a cloud computing environment oriented cognitive intrusion detection method, comprising a data preprocessing unit, a data packet detection unit, a database, a cognitive rule base, an intrusion detection engine unit, an event processing unit, a cloud cognitive inference machine and a statistic analysis unit. A cloud cognitive inference learning module optimizes a feature vector through adoption of a genetic algorithm, so the required training time and monitoring time are shorter than those of other methods; a real-time online detection capability is high; and large-scale parallel computing and large data volume processing capabilities are carried out through full utilization of cloud computing in a cloud computing environment; the operation capability is greatly improved; and a system is relatively safe and efficient.

The invention discloses a clinical decision support system and a decision-making method and belongs to the field of medical artificial intelligence. The clinical decision support system and the decision-making method are provided to solve the problems that in the prior art, existing decisions are made for a special discipline, the correct rate of data diagnosis is still not high, and the resource sharing degree is low. The system comprises a medical knowledge base, an inference machine, an interpreter and a human-computer interaction module providing an interaction platform for patients and doctors. According to the method, after a doctor selects symptom characteristics, the symptom characteristics are input into the internal inference machine, information of a possible disease of a patient and the probability of the disease are calculated, then a doctor client refreshes a new to-be-inquired symptom problem and displays the probability of the disease to the doctor through a chart, and an interpretation module on a page can display interpretation of the possible disease. Through the clinical decision support system and the decision-making method, clinical decision support for primary medical symptoms can be realized, the decision level is increased, and the correct rate of data diagnosis is high.

The invention discloses a fault diagnosis method and expert system for college teaching servers. The method comprises the following steps: acquiring expert knowledge; establishing a fault tree based on the expert knowledge; storing the fault tree in a knowledge base; acquiring fault information a teaching server cluster; storing the fault information in an auxiliary memory pool; comparing the fault tree with the fault information to acquire a server fault processing method; interpreting the fault processing method; and outputting the interpreted fault processing method. The expert system comprises a human-computer interaction module, a fault tree establishment module, a knowledge base, a fault information acquisition module, an auxiliary memory pool, an inference engine and an interpreter. According to the fault diagnosis method and expert system disclosed by the invention, the fault reasons of the servers are accurately discovered by a fault tree analysis method, and an overhaul scheme is provided, thereby overcoming the original inconvenience that restoration cannot be performed in time due to the shortage of IT service personal, greatly reducing the operation costs and the maintenance fees of the college teaching servers and greatly improving the stability and the reliability of the servers.

The invention discloses a water conservation domain information retrieval system and a water conservation domain information retrieval method based on semanteme. The system comprises an information acquisition and storage module for acquiring water conservation subject information from the Internet, storing the water conservation subject information into a water conservation subject information base and constructing an index, a semanteme body module for storing a water conservation domain body and a hownet semanteme body and finishing calculation for similarity of domain vocabularies, and a semanteme query processing module. The method comprises the following steps of: describing domain information by using a fuzzy resource description framework, and constructing the index for the water conservation information by adopting Lucene; constructing a water conservation domain body on the basis of a water conservation official document subject vocabulary table, and performing semanteme expansion on the water conservation professional vocabularies through a Jena inference engine according to the water conservation body to perform the semanteme expansion on general vocabularies on the basis of Hownet; and calculating the relevancy between an expanded vocabulary and a retrieval vocabulary by an improved hownet semanteme vocabulary similarity calculation method. By the system and the method, semanteme expansion is performed on the retrieval vocabulary, and the recall ratio and the precision ratio of the information are improved.

The invention discloses a dependency relationship-based teaching domain ontology automatic generation method. The method comprises the steps of firstly extracting and automatically generating a domainterm list from teaching materials through TF_IDF weights, and converting domain terms in the domain term list into concepts in a teaching domain ontology, thereby generating an initial teaching domain ontology only with a single-layer structure; secondly according to characteristics of various teaching domain relationships, defining dependency relationship-based sentence templates for describingan ontology relationship, thereby forming a sentence template library for describing the ontology relationship; and finally performing template matching, ontology relationship information extraction and ontology relationship automatic generation according to the steps. Apparatuses involved in the method include the domain term list, the sentence template library, a sentence preprocessing module, atemplate inference engine and an ontology relationship generator. The method has the advantages that the templates are not influenced by physical positions of sentence structures; the expandability is high; the teaching domain ontology automatic generation accuracy is high; and the method can be widely applied to automatic construction of the teaching domain ontology in an intelligent teaching system of all disciplines.

The invention discloses a computer assisted instruction (CAI) expert system comprising an expert module, a man-machine interface, a student interface, an inference machine, a regular module, a database and a knowledge base, wherein, the expert module chooses questions from the database for students by the regular module, students feed back the result to the expert module on line, and then the expert module diagnoses the results of the students and chooses knowledge from the database to form the knowledge base for students to study by the regular module; the expert module is a teaching assisted system formed by abundant experiences of educational experts to realize that a virtual teacher formed by abundant experiences of experts carries out teaching on one student; according to on-line feedback, the expert module diagnoses the mark level of the student to call suitable study contents, which fully embodies individual education to students; the called study content can be repeatedly adjusted for several times to perform pertinence interactive teaching and instructing so as to improve studying effect and increase studying interest.

The invention discloses an intelligent household Internet of things control system and method. The system comprises a mobile control terminal and household electrical appliance nodes, wherein the mobile control terminal is internally provided with a visualization rule editor. According to the invention, a rule database is established and is issued to the household intelligent network nodes internally provided with intelligent chips through the visualization rule editor arranged in the mobile control terminal; and when a new event is received, inference machines arranged in the intelligent chips traverse historical event tables in the chips for comparing with the rule database, and in case of successful matching, household electrical appliances are triggered to act, and other nodes are notified through broadcasting. The historical event tables record events occurring in each household electrical appliance node in a household network, the chips automatically record and update historical event tables, the rule database and service sets, different household electrical appliance types are corresponding to the different service sets, and an expert system composed of the inference machines and the rule database endows each household electrical appliance with an intelligent characteristic of triggering different motion according to different conditions, such that the household electrical appliances can still work intelligently in case of disengagement from the Internet and manual control.

The invention discloses a power grid information system fault positioning system and method based on a bidirectional deep neural network. The power grid information system fault positioning system comprises a fault monitoring module, an inference machine, a database and a deep learning module. The method includes; Performing deep neural network forward and reverse training on a fault tree analysisresult; and obtaining a corresponding relation between the fault characteristics and the fault positions, storing the corresponding relation in an expert knowledge base, carrying out power grid information system fault monitoring, writing the fault characteristics obtained according to monitoring data into a characteristic database, and performing reasoning to obtain a fault positioning result according to the expert knowledge base and the fault characteristics. The problem that power grid information system fault positioning is difficult can be effectively solved; The problems of low positioning speed and inaccurate positioning are solved, so that the system can give an alarm at the first time when a fault occurs, intelligent fault positioning and intelligent processing functions of theinformation system can be realized in combination with the fault self-processing system, the operation and maintenance efficiency of the power grid information system can be further improved, and safeand stable operation of the information system is guaranteed.

The invention provides a switch cabinet intelligent monitoring and expert diagnosis system. The system comprises an integral intelligent monitoring device and an expert diagnosis device, wherein the integral intelligent monitoring device collects electric signals, temperature signals and mechanical properties, performs unified processing on the electric signals, temperature signals and mechanical properties, and builds a switch cabinet full life circle condition monitoring information database, the expert diagnosis device is an artificial intelligent system which is built according to switch cabinet professional knowledge and a database of monitoring data, and a switch cabinet fault inference mechanism is used for discovering early fault signals of the switch cabinet.

The invention provides a stress aid decision making experimental device and method based on interaction of multiple sensing channels. The stress aid decision making experimental method based on interaction of the multiple sensing channels comprises the steps that (1) a stress task in a stress running mode is started, an expression signal, a voice signal and a physiological signal of a person to be tested in the stress task are collected, and the collected expression signal, voice signal and physiological signal are fused to obtain a target signal; (2) a stress state type corresponding to the target signal in the stress running mode is obtained according to the target signal; (3) the multichannel reaction priority sequence of the stress state type is determined according to the stress state type, a corresponding stress aid decision making method in a knowledge and method base is used, and the stress decision making of the person to be tested in the stress running mode is assisted according to the stress aid decision making method. According to the stress aid decision making experimental method based on interaction of the multiple sensing channels, the knowledge and method base, a data base and a model base are used comprehensively to design an inference machine, then a stress aid decision making system is established, and man-machine interaction is achieved.

Generative and inference machine learning models with discrete-variable latent spaces are provided. Discrete variables may be transformed by a smoothing transformation with overlapping conditional distributions or made natively reparametrizable by definition over a GUMBEL distribution. Models may be trained by sampling from different models in the positive and negative phase and/or sample with different frequency in the positive and negative phase. Machine learning models may be defined over high-dimensional quantum statistical systems near a phase transition to take advantage of long-range correlations. Machine learning models may be defined over graph-representable input spaces and use multiple spanning trees to form latent representations. Machine learning models may be relaxed via continuous proxies to support a greater range of training techniques, such as importance weighting. Example architectures for (discrete) variational autoencoders using such techniques are also provided. Techniques for improving training efficacy and sparsity of variational autoencoders are also provided.

The invention provides a marine disaster early warning expert system based on data mining. The system comprises 7 modules including a marine data base, a data mining module, a marine disaster knowledge base, a knowledge acquisition interface, an inference machine, an explanation module and a marine disaster early warning display interface. Marine data are abundant and various, large-scale complicated marine data are simplified and compressed by adopting a fuzzy clustering rough set attribute reduction method, attribute factor information related to the marine disaster is retained to the largest extent, information containing a large amount of knowledge is obtained by using the data mining, the information containing the knowledge is regularly extracted and treated, and then strategy rules capable of predicting the marine disasters are found out. The strategy rules organize and form the marine disaster knowledge base and the inference machine, knowledge rules and data in the knowledge base are used for inference, heuristic method inference and judgment are conducted simulating experts in the marine area, marine disasters are evaluated, levels of the marine disasters are judged, and an explanation program is combined with an early warning threshold value to send out disaster early warning towards users in a simple and visual mode. The system effectively solves problems of shortage of expert early warning and marine knowledge personnel on the marine disasters, and provides new ways and methods for popularizing scientific marine disaster early warning technology.

The invention provides an expert system for remote sensing image processing. The expert system for remote sensing image processing comprises a user interface, a system interface, an inference machine, knowledge bases, a management system of the knowledge bases, a database, a database management system, a knowledge obtaining mechanism and an interpretation mechanism. The expert system earnestly and scientifically conclude and summarize and establish the knowledge bases by simulating visual interpretation experiences of a human expert, then quotes the knowledge bases to classify remote sensing data, utilizes a mode recognition method to obtain multiple characteristics of ground feature, provides evidence for remote sensing image interpretation, meanwhile applies an artificial intelligence technology, applies experiences and methods of the remote sensing image interpretation expert, simulates specific thinking process of remote sensing image visual interpretation, and performs remote sensing image interpretation to enable the computer interpretation level to reach the expert level.

The technical scheme of the invention comprises a storage battery pack fault diagnosis system based on big data. The system comprises an interpreter which is used for human-computer interaction; an inference machine which is used for analyzing a storage battery pack fault and performing maintenance; an expert model component which is used for storing storage battery pack configuration and maintenance recommendations; a database which is used for storing historical and real-time data; and an automatic learning machine which is used for automatically learning fault characteristics. The system also comprises a storage battery pack fault diagnosis method based on the big data. The method comprises the steps that instruction transmission is performed on a human-computer interface, the inference machine performs data access and analysis according to instructions, the fault is stored if the fault is a new fault, and the interpreter receives the result and displays the result to a user. The beneficial effects of the storage battery pack fault diagnosis system and method based on the big data are that judgment and analysis are performed according to a model knowledge base by adopting an intelligent summery data, degree of expertise of maintenance personnel is reduced, efficiency is enhanced and thus the system and the method are convenient and rapid.

The invention belongs to the technical field of alloy cast steel rolls, and relates to a method for establishing an alloy cast steel roll material designing expert system on the basis of a neural network. The expert system comprises an integrated database, an inference machine, a knowledge base, a performance optimization module, a man-machine interface and other modules; sample sets for training and testing the neural network are stored in a database of the expert system; knowledge rules of components, processes and properties obtained in a process of training the neural network are stored in the knowledge base in a form of matrix to provide knowledge reserve for inference of the system; the performance optimization module automatically judges performance prediction results and returns to the integrated database of the system; each module of the expert system has a friendly man-machine interface. The expert system constructed by the artificial neural network in the invention has advantages of convenient knowledge maintenance, fast inference and the like, also reduces blindness in an alloy cast steel roll material designing process, and saves time and costs.

The invention discloses a fault diagnosis system based on PDA and a method thereof, wherein a central database server, a network server, client computers and wireless access points are connected by a wired network, the central database server is mounted with a large relation data bank, for establishing a knowledge base and an inference engine; the network server interchanges the information among the central database server and a plurality of PDA; each PDA is accessed into the network via client computers or wireless access points to interchange data with the central database server. The method comprises that users input fault phenomenon; an information pretreatment unit matches a source case with the fault phenomenon, an answer extraction unit extracts and feedbacks a most similar answer according to the inference result to users. The invention provides a system and a method, via which outfield workers can utilize PDA to realize auxiliary maintenance of planes and obtain high fault diagnosis efficiency and ability on PDA.

The invention relates to a method of analysing noumenon inconsistency based on description logic, and belongs to the semantic and noumenon analysis field, which is characterized in building a noumenon warehouse, a buffering noumenon library, a noumenon inconsistency analysis module, a logic inference machine RACER and a noumenon tool Protege. The method analyses inconsistent noumenon stored in the buffering noumenon library via the noumenon inconsistency analysis module and calculates a minimum inconsistent axiomatic set and a minimum inconsistent hitting set of the noumenon to finally gain a minimum error set. The minimum error set is the root of generating the noumenon inconsistency. Deleting or modifying the minimum error set can effectively recover consistency of the noumenon. According to the invention can process to analyse for the noumenon in the situation of semantic inconsistency and find out the root of generating the noumenon inconsistency, so as to effectively recover the consistency of the noumenon.