With the rapid development of artificial intelligence technology and its gradual application in the military field, intelligent radar and intelligent radar technology have attracted wide attention at home and abroad. Strengthening the research of intelligent radar and its key technology is not only the need of radar technology development, but also the key to improve radar combat capability.
1. Introduction
Traditional radar works mainly according to the pre-designed working mode. It has no ability to change the working mode automatically with the change of the target and environment, and it has insufficient adaptability to the target and environment. Although adaptive signal processing technology has been widely used to improve radar anti-clutter, anti-jamming and other aspects of performance, but because adaptive signal processing requires the signal to be processed to satisfy the ideal conditions of stationary characteristics in order to obtain satisfactory performance, so the actual combat environment, especially under complex battlefield conditions adaptive signal processing. Signal processing is also limited. In order to meet the current and future operational requirements, the radar system and technology must be further innovated and developed to continuously enhance the detection capability of radar in order to adapt to increasingly complex operational environment.
The battlefield environment is complex and changeable under the condition of information warfare, which requires that the radar must be able to change its working mode and parameters intelligently according to the requirement of combat mission and actual combat environment, and select countermeasures independently. In order to achieve this, we must add "intelligent" function to the existing radar design, so that the radar can be "intelligent". The rapid development and deep fusion of modern radar technology and artificial intelligence technology will provide strong technical support for the development of "intelligent" radar, i. e. intelligent radar.
2, the basic concept of intelligent radar
(1) AI technology
Up to now, the definition and nature of "intelligence" is still an unsolved problem in scientific research. From the point of view of Engineering technology, intelligence can be understood as "the ability to acquire and store knowledge and use knowledge to solve problems". Its basic characteristics are: perceptual ability; with memory and thinking ability; with learning and adaptive ability; with behavior ability. Artificial intelligence is a subject that studies machine intelligence, that is, using artificial methods and techniques to study intelligent machines or intelligent systems to imitate, extend and expand human intelligence and realize intelligent behavior.
Artificial intelligence is developed on the basis of in-depth study and mutual penetration of computer science, information theory, control energy, system theory, neurophysiology, psychology, linguistics, mathematics, philosophy and other disciplines. It is a highly comprehensive frontier subject and frontier science. It is also a new idea, new theory and new science. Emerging disciplines emerging from technology. Since the advent of artificial intelligence, it has experienced more than 60 years of tortuous development process, has been widely used in many aspects of people's daily life and work, and in the military field has a broad application prospects. At present, artificial intelligence technology has entered a new period of development, artificial intelligence in China has been highly concerned by the state and academia, was written in the 2017 government work report.
(2) the basic concept of intelligent radar
The concept of intelligent radar is in the process of development and formation. The understanding and understanding of intelligent radar will be deepened gradually with the deepening of research. At present, there is no clear definition of intelligent radar at home and abroad. We think that intelligent radar can be understood in broad sense and narrow sense generally. In broad sense, radar using artificial intelligence technology is intelligent radar. The application object of artificial intelligence technology can be part of radar subsystem or single radar system, or radar netting system composed of multiple radars. In a narrow sense, intelligent radar is considered as a kind of "radar with the ability of information acquisition, knowledge learning, autonomous reasoning and decision-making". It can change the transmitting signal form, working mode, processing mode and resource allocation independently according to the external target, environment and task requirements in order to obtain better target detection, tracking and recognition performance. . Compared with the generalized intelligent radar, the narrow sense intelligent radar is an intelligent radar with higher intelligence.
Considering the intelligent coverage, it can be considered that the intelligence of radar subsystem is the primary form of intelligent radar, the intelligence of single radar system is the intermediate form of intelligent radar, and the intelligence of multiple radar netting systems is the advanced form of intelligent radar. Considering the depth of intelligent application, it can be considered that the intelligent radar based on mathematical rules is a primary intelligent radar, the intelligent radar based on empirical knowledge is a secondary intelligent radar, and the intelligent radar based on data-driven is a high-level intelligent radar.
Intelligent radar is the result of deep fusion and comprehensive application of modern radar technology, artificial intelligence technology, communication and information technology, computer network technology and so on. It is an important direction of modern radar development.
3, the development and trend of intelligent radar at home and abroad
(1) overseas research
Since the end of the last century, the Air Force Research Laboratory (AFRL) has been devoted to the research of intelligent signal processing. Its early research mainly focused on constant false alarm processing (ES CFAR) of expert system. Since then, knowledge-based space-time adaptive processing (KB STAP) and digital map information-based space-time adaptive processing (KBMap-STA) have been developed respectively. P) research projects. AFRL and DARPA in the United States have sponsored many research projects, such as knowledge-based radar (KB-Radar), Knowledge-Aided sensor signal processing and expert reasoning (KASSPER), Knowledge-Aided radar (KA-Radar) and automatic intelligent radar system (AIRS). In 2007, the US DARPA also listed KASSPER as one of the main directions of radar technology. In 2015, the United States DARPA launched the "Target Recognition and Adaptation in Competitive Environment" tracking radar target recognition project, using depth learning technology to improve non-cooperative target recognition capabilities.
In the 1980s and 1990s, Guerci conducted a great deal of research in the field of cognitive radio, and put forward the concept of "matched irradiation", which is essentially an intelligent allocation of irradiation power in space. Kershaw and Evans proposed in 1994 to transform the system into a closed-loop radar feedback system by adding an offline link from the receiver to the transmitter. In 2002, the United States Air Force Research Laboratory, the Space and Missile Defense Command and the Navy Research Laboratory jointly organized the Waveform Diversity Task Force to carry out scientific research and experimental work on radar intelligent launching technology, making major breakthroughs in frequency diversity array, space-time coding array and so on.
In 2006, Canadian Simon Haykin first proposed the concept of cognitive radar. Through Knowledge-Aided and adaptive launching, it can realize the continuous interaction and learning with the environment, obtain the information of the environment, combine the prior knowledge and reasoning, constantly adjust the parameters of the radar receiver and transmitter, and self-adaptively detect the target. In 2007, Guerci proposed a knowledge-based cognitive radar system architecture. In 2013, the United States DARPA supported the Radar and Communications Shared Spectrum (SSPARC) research project. Based on cognitive radio and cognitive radar, the communication and radar mutual transmission spectrum usage was realized, and the mutual interference between radar and communication was reduced. In 2016, Greco extended cognitive radar to passive radar.
In addition, knowledge-based decision support, performance evaluation, simulation training, fault diagnosis technology are also important aspects of intelligent radar technology research, and very early began to become practical. In 1991, the United States Department of Defense formally issued the "Comprehensive Diagnosis" standard. As the main way to improve the operational readiness of the new generation weapon system and reduce the cost of support, fault diagnosis using artificial intelligence technology has been applied to the new research radar system.
Generally speaking, although the concept of "smart radar" is seldom mentioned abroad, a lot of research work has been carried out in the field of smart radar related technology, and the technology maturity has been continuously improved. Many research results are on the eve of practical application, and some of the research has become practical.
(2) domestic research situation
Since 2008, a number of domestic units have carried out research in the field of cognitive radar, including environmental perception and description, optimal waveform design, adaptive signal processing and so on. At present, there is a certain research foundation in the field of intelligent radar information processing and clutter suppression technology in China. The main research features are combining MIMO radar, cognitive radar, software radar and other research directions, focusing on the formation of knowledge on the basis of statistical analysis of target and environment data, and further using multi-waveform optimization. Adaptive processing and cognitive structure with feedback can improve the detection performance of radar system for weak targets, clutter suppression performance and the utilization efficiency of radar system resources. In the research of intelligent radar target recognition technology, there is also a more extensive research foundation in China, including the use of neural networks, support vector machines to classify and recognize the results of radar range profiles, classification and recognition of SAR images. In addition, in recent years, China has also paid close attention to cognitive electronic warfare, and carried out the preliminary research work, which to a large extent promotes the research work of radar intelligent anti-jamming technology.
Overall, China has been following up and carrying out some research in the field of smart radar technology, but there is still a big gap compared with foreign countries in the aspects of systematicness, advancement and innovation. The system structure and technology system have not been established.
(3) development trend of intelligent radar
Deep intelligence. With the development of artificial intelligence technology, the degree of radar intelligence is bound to deepen. At present, adaptive algorithm processing technology based on mathematical rules has been widely used. Radar begins to have primary intelligence. Advanced intelligent radar based on depth learning will gradually appear in the future.
Widely practiced. Just as the emergence of smart phones makes traditional mobile phones disappear quickly, smart radar will be widely used in various combat platforms. Cognitive radar and Cognitive Electronic Warfare (CEW) follow each other like shadows, constantly advancing in contradictions, intelligent radar technology system itself will gradually mature through continuous learning.
Multi function integration. Intelligence often means multi-function. With the development of active phased array technology, digital array technology and software radio technology, multi-function integrated intelligent radar, which integrates the functions of air defense, anti-missile and anti-defense, will appear soon.
Integrated RF integration. Detection, reconnaissance, interference, communication, attack, control and evaluation are important elements of the future information war, but the underlying technical mechanism is extremely compatible. In the future, it will be the inevitable development trend of military electronic information technology to realize the above-mentioned integrated radio frequency integration based on intelligent radar technology under strong confrontation conditions and complex electromagnetic environment.
Self organizing networking. Networking is undoubtedly an important direction of future radar development. There are two forms of future intelligent radar self-organizing network: one is distributed, each radar has a higher intelligence; the other is centralized, that is, to set up a central base station, only the base station radar has a higher intelligence.
4, the key technology of intelligent radar
The goal of the development of intelligent radar is to enable the radar to have the ability of intelligent perception of environment, processing information, decision-making response, control action and other independent behavior. In the process of promoting the evolution of intelligent radar from primary form to advanced form, from primary intelligence to advanced intelligence, the following smart radar gates should be broken through. Key technology to promote the continuous development of intelligent radar.
(1) intelligent information processing technology
Aiming at the targets with low radar cross section such as stealth aircraft and the complex ground/sea clutter environment, this paper introduces the technical thinking of artificial intelligence, uses various prior knowledge to form the knowledge base of the target and environment, and adjusts independently through the process of machine learning, analysis and judgment on the basis of continuously obtaining the target and environment characteristic information. The frequency, waveform and energy of radar system can improve the performance of detecting and tracking weak targets. The main research contents include target detection and tracking technology based on prior knowledge, radar information processing knowledge formation, rule extraction and logical reasoning technology.
(2) intelligent clutter suppression technology
Aiming at the characteristics and difficult problems of strong clutter and high space-time coupling in the down-looking working environment of airborne early warning radar, a priori knowledge of clutter environment is formed by continuous knowledge accumulation and learning based on artificial intelligence theory. Combining with advanced multi-waveform flexible transmitting technology of digital array radar, the clutter suppression performance and complexity are improved. Moving target detection performance in clutter environment. The main research contents include clutter environment perception and parametric characterization technology, radar emission pattern and waveform optimization technology.
(3) intelligent anti-jamming technology
Aiming at the characteristics of non-stationary and high dynamic changes of jamming signals in electronic warfare environment and the development of cognitive electronic warfare technology, this paper adopts the theory and technology of artificial intelligence to accomplish cognition and decision-making quickly on the basis of real-time perception and learning of electronic warfare environment, and multi-dimension such as time, space, frequency, waveform, polarization and so on. Resources evaluation and scheduling are carried out to achieve intelligent game under complex interference conditions. The main research contents include jamming environment sensing technology, anti-jamming measures automatic optimization selection technology, anti-jamming effect feedback and evaluation technology.
(4) intelligent target recognition technology
Aiming at the classification and recognition requirements of target types, attributes and quantities in actual combat environment, the machine learning theories and methods such as depth learning and support vector machine are adopted to complete the tasks of target and environment feature extraction and pattern recognition, and realize the effective recognition of non-cooperative targets. The main research contents include target feature learning and recognition method based on depth network, target classification and recognition method based on large data and small samples, man-machine collaborative target recognition method and so on.
(5) intelligent resource allocation and scheduling technology
Aiming at the complex combat environment of radar, such as multiple targets, true/false warheads, active/passive jamming and so on, the complex scene discrimination calculation, fast intelligent decision-making and flexible resource allocation are realized by means of artificial intelligence. The anti-jamming of real-time implementation system, reasonable allocation of target tracking resources and effective implementation of key targets are solved. Identify complex problems and make full use of existing radar resources to improve radar operational effectiveness. The main research contents include radar resource intelligent dynamic scheduling and task planning technology, radar resource intelligent dynamic reconstruction and real-time evaluation technology.
(6) intelligent multi-source data mining technology
Aiming at the requirement of improving command and decision-making ability of early warning operation, this paper adopts the theory and method of artificial intelligence to organize, classify, analyze and interpret the large amount of data accumulated in the peacetime combat readiness, training and test of radar system in order to accumulatively describe the information in the data, and adaptively learns various numbers from different sources through artificial intelligence. According to the data, the related information is found, and the multi-dimensional information content is formed. Further experiments and operations are carried out to improve the intelligent support ability for early warning command and decision-making. The main research contents include knowledge model based on multi-radar data, knowledge acquisition, knowledge organization, knowledge-based learning algorithm, reasoning analysis, and distributed intelligent radar network expert system and decision support system.
5, concluding remarks
At present, the development of radar equipment and technology is faced with many major challenges. What is the future of radar? How to develop future radar is a question that must be seriously considered and answered before us. The development of radar equipment and technology must not be "closed-door". It must follow the trend of the development of advanced technology, merge new technology and speed up the pace of innovation and development. Artificial intelligence is a new frontier science which is developing rapidly. It is another revolution after the three industrial revolutions. With the development of big data and cloud computing technology, artificial intelligence technology such as natural language understanding, human-computer interaction, intelligent decision-making control and so on may have great application breakthroughs in radar. Smart radar is a new concept of radar system. It is also an important direction of radar technology development in the future. Although the basic concepts and key technologies of intelligent radar are not mature, and intelligent radar will face many great challenges from the engineering implementation, the future development of intelligent radar should arouse our great concern. To adapt to the development trend of intelligent perception, the deep integration of radar technology and artificial intelligence technology will make the future radar more intelligent, thus bringing a bright future to the future of radar.