ZHEJIANG DEYE AUTOMOBILE PARTS CO.,LTD , https://www.deyemotor.com
Artificial neural network is the basic technology in today's intelligent technology. Its connection mechanism is tied with the symbolic reasoning mechanism of artificial intelligence. It has become the two major camps of intelligent technology. It simulates the anatomy and physiology of the human brain. It uses a number of parallel simple neurons to form a network with certain topological structures. It accepts external information and stimulates each other. It is better at distributed storage, associative memory, feedback refinement, and black box mapping. Balanced weights, dynamic approximations, holographic records, fault tolerance, and a large number of neuron interconnects create strong self-learning, adaptive, self-organizing, self-diagnosis, and self-healing capabilities. Continuous feedback of intensity, dynamic analysis, close cooperation with language, audio-visual human-computer interface, automatic knowledge and experience of human experts can be automatically obtained, and the logical reasoning, image thinking and even inspiration of the human brain can be simulated, and various inaccuracies can be properly handled. , imperfect, uncertain information, reasoning to draw correct conclusions.
Fuzzy logic imitates the uncertainty of the human brain to determine the concept, reasoning thinking mode, for the unknown or uncertain model of the model description system, and the strong nonlinear, large lag control object, the use of fuzzy sets and fuzzy rules to reason, express transitional boundaries Or qualitative knowledge experience, simulating the way of human brain, applying fuzzy comprehensive judgment, inference to solve the regular fuzzy information problem that conventional methods are difficult to deal with. Fuzzy logic is good at expressing the unclear qualitative knowledge and experience. It uses the concept of membership function to distinguish fuzzy sets, handle fuzzy relationships, simulate the human brain to implement rule-based reasoning, and solve the various problems caused by the logic breaking of "the Law of Exclusion Center". Determine the problem.
The genetic algorithm is a search method that suppresses the computational explosion of the search space with the feature of “electron beam searchâ€. It can fully search in multiple points of the solution space, use genetic algorithms, repeatedly cross, and operate in abrupt manner to simulate the interior of things. The diversity and adaptability to environmental changes are characterized by their strong operability and ability to avoid falling into local minima at the same time, allowing the problem to converge globally quickly. This is a class of self-disciplined decentralized systems that can utilize multiple information globally. Evolvable hardware (EHW) made by using evolutionary methods such as genetic algorithm (GA) can generate novel circuits that exceed the technical synthesis and designer capabilities of existing models, especially GA's unique global optimization performance, enabling it to self-learn, Adaptive, self-organizing, and self-evolutioning capabilities are more fully utilized. Automated synthesis, expansion of massively parallel processing (MPP) in unmanned space, and real-time, flexible configuration and call-up of function-level EHWs based on EPGA solve multidimensional problems. The complex problems of uncertainty in space opened the course.
The expert system collects the knowledge and experience of the application of human experts and imitates how experts handle knowledge and solve problems. It is compiled into a computer intelligence software system, and in the case of continuously obtaining feedback information through man-machine integration, the rules, examples and A problem solving or control system where the model implements independent decision making. This kind of computer intelligence system is inspiring, transparent and flexible. It is efficient, accurate, comprehensive, and fast and tireless to complete its work without the influence of time, space and environment. It has the ability to solve problems and knowledge. Extensiveness can surpass human experts and overcome the errors and errors caused by human experts' negligence, forgetfulness, tension, tiredness and other interference factors. Therefore, its promotion and application have enormous economic and social benefits.
Pattern recognition is an intelligent decision method and technology that simulates the human brain's image thinking, identifies, determines, and processes things according to the characteristics, images, or relationships of things. It is widely used in scientific research and production, and it is a valuable technological method. .
Rough set theory is to process discretely normalized datasets obtained in the measurement. Through algebraic operations based on indiscernible relations of set elements, use of a large number of useful features and effective data in conditional and result attributes to discover knowledge in decision-making. The initial simplified calculation of the rules obtains the nuclear value, and then further simplifies the rules and selects the minimum decision algorithm according to the problem requirements for practical application, removing redundant attributes in a large amount of information and reducing the number of dimensions and attributes of the information space. It can greatly simplify the network structure and the number of samples and shorten the training time. It is a fundamental analysis method in intelligent technology. This method is based on the measurement data set and acquires knowledge, so it is of great significance to the intelligent development of virtual instruments.
Chaotic movement is a highly unstable movement confined to a finite phase space in a deterministic system. It is disorderly ordering, and it causes things to show some confusion on the surface in long-term behavior. Chaotic phenomena are characterized by "hidden ordering behind aperiodic" and "sensitive dependence on initial conditions", making full use of chaotic features, implementing nonlinear decision-making and prediction, nonlinear system identification, and pattern recognition in intelligent information processing. , image data compression, high-performance confidentiality, multi-target search, and infinitely rich, fantastic computer painting and other magical applications.
Fractal theory studies the proportional self-similarity of non-smooth and non-differentiable geometric shapes generated by nonlinear systems and their intrinsic structures. It provides powerful tools and methods for studying the law of movement of all complex things in nature.
Wavelet analysis is the backbone and the perfect crystallization of the modern analysis of mathematics. From the intuitive visual point of view, wavelets refer to the shortest, simplest, and the same attenuation waves that people can observe; mathematically, the wavelet function f(t) has its center three conditions. Window function, which not only can characterize the localization of the signal in the time domain and frequency domain, but also can completely retain all the information of the signal. It also has the nature of zooming, that is, it has a very narrow time for high-frequency signals that appear only instantaneously. The window, but also has a wide range of different scale transforms at low frequencies. The essence of wavelet analysis is to reflect the duality of wave-particle duality in the world of things and the dialectical relationship between local and overall multi-level display. Its most attractive characteristics lie in time-frequency positioning and multi-scale approximation capabilities, in adaptive control, robust control, and non- Linear control, process identification, neural networks and other fields have achieved fruitful results.
Fractal and chaos are two aspects that are essentially the same. Chaotic events show similar patterns of change at different times, while fractals are similar in the spatial scale. Chaos is concerned with its complex process of instability, divergence, and convergence, while fractals are intuitive geometric languages ​​that characterize chaotic motion. The organic combination of chaos, fractal and wavelet analysis has a rich connotation and profound philosophy. It will certainly provide powerful tools for the resolution of major microscopic technological problems such as automatic assembly of material molecules, high-speed gene sequencing, and prediction of efficient protein structure. It will also open up a bright future for instrumentation, virtualization, networking and intelligence.
The matter-element extension method is based on the comparison and optimization of a variety of known general decisions, and based on the needs of inconsistent contradictions that are generated at each level and in each stage, it can be used to break through conventionally and expandly to take creative decisions. Skills, seize the key strategy, maximize the satisfaction of the main system, the contradictions of incompatibility into compatible relationships, so as to achieve the best global decision-making goals. It is a powerful means to resolve minor contradictions in complex systems and resolve major contradictions and key problems. It will also make significant contributions to the development of instrumentation, virtualization, networking, and intelligence.
The data fusion technology is a kind of technical method that measures the data of multiple information sources, assigns different weights and weights according to its importance and credibility in the whole system, and comprehensively calculates the overall optimal representation of the characteristic attributes. It is an optimized measurement and characterization technique for the properties of complex things, which is of great significance for high-tech development research.
In short, the smart technologies in today's world are developing rapidly and comprehensively.
Second, the application of intelligent technology in instrumentation and measurement The application of intelligent automation technology is infiltrating the instrumentation industry.
(1) Application in Instrumentation Structure and Performance Improvement First, intelligent automation technology opens up broad prospects for the application of instrumentation and measurement in related fields. Intelligent hardware and software are used to enable each instrument or instrument to accurately analyze and process current and previous data information at any time. The measurement process is appropriately abstracted from low, medium, and high levels to improve existing measurement systems. The performance and efficiency of the extended traditional measurement system, such as the use of neural networks, genetic algorithms, evolutionary computing, chaos control and other intelligent technologies, make instruments and meters to achieve high-speed, high efficiency, multi-function, high mobility and other performance.
Secondly, microchip technologies such as microprocessors and microcontrollers can also be used in different instruments of distributed systems to design fuzzy control programs, set critical values ​​of various measurement data, and apply fuzzy rules to fuzzy logic inference techniques. Various fuzzy relationships make various types of fuzzy decisions. The advantage is that it is not necessary to establish a mathematical model of the controlled object, nor does it require a large amount of test data. Simply follow the experience, summarize the appropriate control rules, and use off-line calculation and on-site debugging of the application chip to generate accurate results according to our needs and accuracy. Analysis and punctual control actions.
Especially in sensor measurement, the application of intelligent automation technology is more extensive. Using software to implement signal filtering, such as fast Fourier transform, short-time Fourier transform, wavelet transform and other technologies, is an effective way to simplify the hardware, improve the signal-to-noise ratio, and improve the dynamic characteristics of the sensor, but it needs to determine the dynamic mathematical model of the sensor, and the higher order The filter has poor real-time performance. Using neural network technology, high-performance autocorrelation filtering and adaptive filtering can be realized. Make full use of the powerful self-learning, self-adaptive and self-organizing capabilities of artificial neural network technology, association and memory functions, and the characteristics of black box mapping between input and output of nonlinear complex relationships, in terms of applicability and fast real-time performance, etc. All of them will greatly exceed the complex function type, and can make full use of multi-sensor resources to comprehensively obtain more accurate and more reliable conclusions. Among them, real-time and non-real-time, fast-changing and gradual-change, fuzzy, and deterministic data information may support each other, and may also conflict with each other. At this time, the object features are extracted and integrated until the final decision is made, and correct judgments are made. Will become a difficult point. So neural network or fuzzy logic will be the most worthy method. For example, gas sensor arrays are used for gas mixture identification. In signal processing methods, self-organizing map networks and BP networks can be combined to classify and then identify components, and the full-scale fitting of traditional methods can be converted into subsections. In order to reduce the complexity of the algorithm and improve the recognition rate. Another example is the difficulty of detecting and identifying food taste signals, which was once the major obstacle for research and development organizations. Now it is possible to use wavelet transforms for data compression and feature extraction, and then input the data into a fuzzy neural network trained with genetic algorithms, which greatly improves the recognition rate for simple compound tastes. Another example is the assessment of the quality of fabrics, the handling of tactile signals by flexible operators, and the field of fault diagnosis in machines. Intelligent automation technology has also achieved numerous successful examples.
(2) The combination of application instrumentation, measurement technology, and computer technology in the design of virtual instrument structures not only greatly improves the measurement accuracy and the level of intelligent automation, especially the rapid development of virtual instrumentation in which the computer's hardware is softened and the software is modularized. As well as its integration with networked system resource programs and optimized performance configuration, it has created more and more favorable conditions for the rapid improvement of the intelligent level of instrumentation.
In instrument and meter structure design, instrument manufacturers used to provide users with intelligent virtual instrument plug-and-play instrument drivers in the form of source code. In order to simplify the end user's operation and development process, they continuously improve the operating efficiency, and the programming quality. With programming flexibility, relevant instrument manufacturers have made a new set of intelligent instrument driver software specifications based on VXI plug-and-play bus instrument driver standards, and have made various improvements in the following aspects of virtual instrument structure and performance.
First, consider a high-level programming interface that takes into account the user's intuitive, easy-to-use, and maximum operating efficiency, and maintains the original VXI bus plug-and-play standard, to provide the same functional function call format.
Secondly, based on the latest Labwindows/CVI5.0 built-in development tools, intelligent instrumentation is used to enable IVI instrument driver code to be automatically generated under human-computer interaction. This simplifies a lot of programming. The workload also unified the programming structure and style of the driver code, and also greatly facilitated the use and maintenance of users at different levels.
Once again, a series of smart methods are applied to identify, track and manage all the various instrument states and settings so that the user can directly enter all low-level settings, and through intelligent state management, so that users can, according to their needs, “test development†and “normal Run "switch between the two modes at will. In Test Development mode, the drive can intelligently automate a series of status checks to help detect various programming errors. When the program is debugged and put into use normally, the user can switch to the "normal operation" mode so that the driver software can run at a high speed. This will not only ensure the safety and reliability of the instrument, but also allow the software to be put into high-speed operation at any time and maximize its operating efficiency.
In addition, because of adopting various intelligent methods, the driver can realize multi-threading and safe operation at the same time, and perform multi-thread parallel test; at the same time, the driver also has powerful simulation function, and can be developed and tested without connecting actual instruments. program.
The last characteristic is that the driver operation is only related to the test function, and has nothing to do with the interface bus mode adopted by the instrument. Only an initialization function InitwithOptions is used to distinguish between the instrument interface bus and the regional heterogeneity.
In short, because the virtual instrument adopts a series of intelligent and automated means, it has completely changed the operating efficiency of the VXI bus plug-and-play standard instrument driver, the programming structure, style is inconsistent, programming is difficult, the quality is low, and the workload is large. Maintenance troubles and a series of defects, thus achieving a comprehensive and unified operation under conditions of high efficiency, high quality, safety, reliability, ease of use, and flexibility, shows the profound impact of intelligent automation technology on the rapid development of virtual instruments and the entire instrumentation industry.
(3) Application of Instrument and Instrument Networking Since the instrument and computer once form a network, they can use their intelligent hardware and software (such as pattern recognition, neural network self-learning, self-adaptation, self-organizing, and associative memory functions) to achieve full flexibility. Invokes and reasonably configures the respective resource characteristics and potential of various computers and instrumentation on the Internet, resulting in a combined advantage of 1+1>2. For example, digital multimeters and oscilloscopes connected to the Web can now be used to distinguish different time and space conditions and category characteristics of instrumentation and to measure critical values ​​using a digital multimeter and oscilloscope connected to the Web to make different characteristic responses; it is also possible to use distributed data. The acquisition system replaces the data acquisition equipment that was used alone in the past to implement remote measurement and data collection across Ethernet or other networks, and storage and application of classification.
The networked intelligent measurement environment organically links various types of computers and instruments of different tasks on the Internet to complete various forms of tasks, such as sending data to a place where it is needed, after collecting data at a certain place. Copy multiple copies of the same data on demand, and send them to various departments in need; or send the measurement results to a remote database regularly for storage at any time when needed. Multiple users can monitor the same process at the same time. For example, engineering and technical personnel, quality control personnel, and supervisory leaders of each department can simultaneously monitor and control the same production and transportation process at remote and remote locations without having to visit the site. From all aspects of data, make decisions or establish databases to analyze the laws of phenomena. In the event of a problem, immediate or reconfiguration can be immediately demonstrated, or decisions can be immediately discussed and immediate action taken.
In addition, intelligent reconstruction information processing technology will also create a broader stage for instrumentation. Reconfigurable computers that combine the advantages of computers and application-specific integrated circuits (ASICs) not only require flexible configuration of a large number of programmable logic cell arrays (FPGAs) according to different computational tasks, but also their instruction level, bit level, and pipeline level. As well as task-level parallel computing, it runs hundreds of times faster than general-purpose computers.
To sum up, with the increasingly deep application of intelligent automation technology and the continuous expansion of application scope and scale, the development level of China's instrumentation industry will surely move to a higher level.
III. Future prospects of intelligent automation of instrumentation The application of intelligent technology in instrumentation is rapidly developing with each passing day, and new technologies in many other fields are continuously blending in. For example, on the basis of giving full play to the highest-speed physical properties of the photo-electric beam current, intelligence is increasingly turning to human brains. Actively use the human brain mechanism and the organic intelligence of the biological DNA chip, combined with the high-efficiency and dynamic advantages of the inorganic intelligence of electrons and photon counting speeds, and make the material intelligent, and then interact with virtualization to improve. Nowadays, optical interconnect technology is overcoming a physical limit of the physical nature of electrical interconnect technology with a series of unique physical properties such as extremely high space-time bandwidth, minimal electromagnetic interference, and small interconnect power consumption. The flexible, high-speed, real-time reconstruction of the network interconnection structure greatly improves parallel processing capabilities and creates a new world. This will create a solid foundation for mankind to create a variety of human-computer integration systems and colorful humanoid high-efficiency and high-efficiency automation systems, so as to continuously push human productivity into a new higher realm and make human life towards the smart world. Happy and beautiful tomorrow strides forward!
First, the smart technology to fully revitalize the current branch of intelligent technology, vigorous and prosperous, has been rapid development at home and abroad, such as fuzzy logic, genetic algorithms, neural networks, expert systems, human-like intelligence, rough set theory, matter-element extension Methods, knowledge engineering, pattern recognition, qualitative control, wavelet analysis, fractal geometry, chaos control, data fusion technology, and so on, can be said to be the Eight Immortals crossing the sea, each showing supernatural powers. Each has its own strengths, combined, complement each other and complement each other.