【Author: Northwestern Polytechnical University】
Abstract: Function allocation is an important content in the design process of complex human-machine intelligent systems. It requires the application of systematic analysis methods to reasonably allocate tasks between humans and machines and scientifically design the functional combination of the two. This paper analyzes the research status and existing problems . Aiming at the design requirements of complex human-machine intelligent systems, the research direction of system function allocation method is pointed out.
1 Introduction
In a complex intelligent human-machine system, there are two intelligent individuals —human and intelligent machine. Such a man-machine system is very different from the traditional man-machine system in terms of composition and internal mechanism. It is an intelligent system that combines man-machine and has the characteristics of intelligence, openness, complexity, and emergence. In order to study this kind of system, the famous American scholars Lenat and Feigenbaum put forward the concept of Man-Machine Intelligent System [1], which is similar to the concept of Human-Automation proposed by Sheridan [1]. 2] etc. Qian Xuesen, a famous scientist in China, and Dai Ruwei from the Chinese Academy of Sciences put forward the idea of man-machine integrated integration [3, 4]. Lu Yongxiang and Chen Ying of Zhejiang University put forward the theory of man-machine integration [5, 6]. Long Shengzhao and others put forward the human-machine-environment system engineering theory [7], which provides a lot of theoretical basis for the study of human-machine intelligent system. Although there is no unified and rigorous theoretical system for the study of complex human-computer intelligence systems, common points can be found from the above methods, that is, the whole life cycle process from the beginning of the system design to the system implementation can be found. Considering the relationship between humans and machines, the rational distribution of human-machine functions is the key to embodying the intelligence of complex human-machine systems.
2 The concept of function allocation
As early as 1951, Fitts first clearly proposed the concept of function allocation [8]. It refers to the process of assigning humans or machines. This kind of function allocation activity occurs in the early stage of the system synthesis and evaluation process, so it belongs to the research category of systems engineering. The function allocation here mainly emphasizes the function decomposition among the system components. Later, due to the development of machinery, automation and computer technology, most of the systems belong to human-machine systems. The function needs to take into account more and more human factors in the system, so it is specially used as a system. An important research content called human-machine function allocation. Since the concept proposed, it has been widely used in the field of industry and automation, especially in the design of complex systems such as nuclear power plant monitoring automation systems, air traffic control systems, aircraft cockpits , and manned spacecraft .
3. Research status of function allocation method
3.1 Research situation abroad
The principle of function allocation is combined with a certain application background, and various function allocation methods are gradually formed. At present, several influential methods of function allocation in the world are: human-machine ability comparison allocation method, Price decision diagram method, Sheffield method, automatic classification and hierarchical design method, York method and so on.
3.1.1 Allocation method of human-machine capability comparison
It is the original function distribution method, such as the famous Fitts Lists distribution method, and it is also the most commonly used method in the early simple industrial automation monitoring systems . Table 1 shows the respective advantages , also known as MABA-MABA method.
3.1.2 Price decision diagram method
The Price decision diagram method [9] compares any function from the characteristics of both human and machine, and then makes an evaluation based on efficiency, speed, reliability , technical feasibility, etc. The evaluation result is a complex value (human performance value is the real part, and the performance value of the machine is the imaginary part). This complex value falls in a certain region of the decision diagram, as shown in Figure 1.
Figure 1 Price decision diagram
The decision diagram consists of 6 regions, each of which corresponds to a different human-machine performance and allocation scheme: ① means assigning functions to machines; ② means assigning functions to machines; ③ means assigning humans and machines. There is an optimal distribution point for machines; ④ assign functions to people; ⑤ assign functions to people; ⑥ use other methods to redesign.
Although the Price decision method further clarifies the process of human-machine function allocation based on the FittsLists allocation method , it does not have a clear description of how to calculate performance, and it is objectively difficult to calculate the performance of humans and machines.
3.1.3 Sheffield法
The Sheffield method [10] is a function allocation method developed by the University of Sheffield in the United Kingdom when designing the naval ship control system . It needs to consider more than 100 decision-making criteria in the allocation process , which are divided into 8 groups, which not only consider the ability characteristics of human operation design, sociality, training, Factors such as safety, in addition to the accuracy and cost of automation. Its main process is shown in Figure 2.
Figure 2 Sheffield method flow chart
The advantage of Sheffield method is that the factors considered are comprehensive, and it includes the static and dynamic function allocation process of the system. It is mainly designed for a naval ship control system, so it also considers the distribution of functions among ship operators. But it also has obvious deficiencies. First, because there are too many , the design task cannot be operated due to lack of relevant information; secondly, because the Sheffield method must decompose the functions to be able completely assigned to people or machines, that is, to be fine enough. Granularity to implement operations, but this is often not possible in a complex system.
3.1.4 Automatic classification and hierarchical design method
This method was proposed by British scientists Parasuraman and Sheridan [11], and is mainly used in industrial automation systems such as nuclear power plant monitoring. This method believes that the working process of any human-machine automation system is similar to human information processing, which can be divided into four steps : acquisition, analysis, decision-making, and action. The degree of automation of the machine is divided into 10 consecutive levels as shown in Table 2.
On this basis, the system functions are classified according to the above four steps, and the automation degree of the functions belonging to each classification is determined , and then multi-level evaluation criteria are established to gradually modify the distribution results until it is finally determined that the system should be adopted. type and level of automation. Its flow chart is shown in 3.
Figure 3 Flowchart of automatic classification and level design
3.1.5 York method
It is a scenario-based (Scenario) function allocation method proposed by Dearden et al. from York University in the United Kingdom [12]. It was originally developed for the design of naval ships. Due to the relatively good results, it was later used. Successfully used in the design of the function . The basic steps of the York method are shown in Figure 4.
Figure 4 York method flow chart
This design method also takes into account the environmental factors when the function is running , so it is a relatively complete function allocation method. But it doesn't take into account the distribution of functions among the people in the system.
3.2 Domestic research situation
Domestic research on function allocation is still in its infancy. Zhang Wei from Western University of Technology and others studied the dynamic distribution of human-machine functions [13], and proposed the principles and methods of human-machine function distribution in UAV ground control systems, mainly for remote UAV monitoring systems . , the factors considered in the function allocation are also
It is relatively simple and is not suitable for the design and research of complex systems. Zhou Jiaxiang et al. studied [14], which provided a certain basis for the optimal design of manned spacecraft's optimal function allocation. However, some initial input difficult to obtain, which has a direct impact on the final assignment results.
influence.
4 existing problems
At present, in the system function allocation method and its application, the main problems are as follows [15]:
(1) The generality of the allocation method is poor. The application range of extremely wide. In their respective fields, due to the different application environments, task nature and technology involved, and the allocation standards are not uniform, the function allocation methods in various fields cannot be well compatible with each other. Shared use greatly limits the application of function allocation in engineering.
(2) The allocation standard is single, and the allocation process is relatively simple. In practical engineering applications, function allocation is often the most easily overlooked link. Even if function allocation is considered at the beginning of the design, only a single criterion, such as system load or cost, is selected. The simplification of distribution standards will inevitably lead to the simplification of the distribution process. In this case, the system designed may cause a single index of the whole system to be high, while other indicators are low, so the comprehensive performance often fails to meet the design standards.
(3) The function allocation process and the design process are not closely integrated , there is no engineering method, and there is not enough consideration for environmental factors. The traditional function allocation method considers function allocation as a separate process, which is not closely integrated with system engineering design, and requires the participation of function allocation experts, and it is difficult for ordinary designers to participate, which causes The disconnect between function allocation and engineering design increases design costs. In addition, due to the lack of sufficient consideration of environmental factors, when the system is put into use, the system does not have sufficient dynamic adjustment capability , and may cause the system to crash or fail.
5 Research directions of function allocation methods for complex human-machine intelligent systems
The most important feature of complex human-machine intelligence system is that there are three levels of information coupling between perception, decision-making and execution. It requires that the functions of human and machine can be adjusted dynamically different degrees of intelligence can be reflected in different stages of the task. Therefore, it is difficult to apply a single function allocation standard and method to meet the design requirements of complex human-machine intelligence systems. According to the existing methods, the following suggestions are put forward for the function allocation method of complex human-machine intelligent system :
(1) The characteristics of people and machines in the system must be analyzed to determine the influencing factors of function allocation. The "machine" in the complex human-computer intelligence system is a broad concept, including all kinds of software and hardware facilities other than humans, such as machinery, automation, and computers related to system tasks . The people in the system should refer to a group of people, first of all, he has the characteristics of physiological limitations of ordinary people; secondly, because of training and learning, he has the characteristics of professionals; finally, there are individual differences between . Therefore, it is necessary to focus on the analysis of human characteristics . On the basis of the analysis of human and machine characteristics, the factors that affect the allocation of functions are selected, that is, specific constraints , such as reliability, cost, performance, etc. for machines, and workload and psychological pressure for humans. , physiological limits, reliability, etc.
(2) Before the function allocation, the function analysis must be carried out in combination with the system task scene, which is the basis of all subsequent allocation activities. The purpose of functional analysis is to identify specific design criteria , which break down the system-level requirements into subsystems, and the hierarchy needed to identify the input design criteria and/or constraints on various elements of the system . Ultimately we were able to determine the set of objects for functional assignment in each task scenario. Obviously, different task scenarios have different sets of function assignment objects. For an intelligent system, each function should be classified and the possible intelligence level of each type of function should be determined, and their combination constitutes the basic scheme of function allocation.
(3) Function allocation includes both static function allocation and more importantly, dynamic function allocation process. In the static function assignment stage, there is a lack of necessary initial data related to assignment criteria (these data are used to evaluate the performance ), and the real environment of the task cannot be constructed. Therefore, some pre-estimation methods can only be adopted , making predictions of human and machine performance and other influencing factors based on the assumptions of each allocation scheme in a specific task environment , and then evaluating which scheme is most suitable , and then describing the current situation at that time. Human-machine relationship. In the dynamic allocation stage , the pros and cons of the function allocation scheme can be evaluated by means of simulation methods or prototype systems and by means of human-machine performance experiments, and modifications and adjustments can be made .
6 Conclusion
After decades of development, the theory and method of function allocation have achieved fruitful and effective results in the field of industrial automation. With the advancement of science and technology, automation systems are gradually developing into complex human-machine intelligent systems, which puts forward higher requirements for the design and synthesis of the system, especially for the function allocation of the system . For this reason, system researchers must design, analyze, model and optimize using systems engineering thinking, mathematical methods, and modeling theory from the perspective of systems theory, and implement human-machine functions. Reasonable distribution, make full use of the respective , so as to establish a truly complex man-machine intelligent system.
references:
[1]LenatDB,FeigenbaumEA.OntheThresholdofKnowledge[J] ArtificialIntelligence, 1991, 13(1):3 -10.
[2] SheridanTB, translated by Hu Baosheng. Human and Automation——System Design and Research Issues [M]. Xi'an: Xi'an Jiaotong University Press, 2007.
[3] Wang Shouyun, Yu Jingyuan, Dai Ruwei, et al. Open complex giant system [M]. Hangzhou: Zhejiang Science and Technology Press, 1996.
[4] Qian Xuesen, Yu Jingyuan, Dai Ruwei. A new field of science—— Open complex giant system and its methodology [J]. Nature Journal, 1999, 13(1):3-10.
[ 5] Lu Yongxiang. On Human-Machine Integration System and Technology [J]. Chinese Journal of Mechanical Engineering , 1994, 30(6):1-9.
[6] Chen Ying, Yang Canjun. Theory and Method of Human-Machine Intelligent System [M]. Hangzhou : Zhejiang University Press, 2006.
[7] Long Shengzhao, Huang Duansheng, Chen Daomu, et al. Human-Machine-Environmental Systems Engineering Theory and Application Basis [M]. Beijing: Science Press, 2004.
[8]FittsPM.HumanEngineeringforanEffectiveAir NavigationandTrafficControlSystem[M] .Washington,DC:NationalResearchCouncil, 1951.
[9]PriceHE.TheAllocationofFunctionsinSystems[J].HumanFactors,1985,27(1):33 -45.
[10] OlderM, CleggC, WatersonP.ReportontheRevisedMethodofFunctionAllocationandItsPreliminaryEvaluation [ R] InstituteofWorkPsychology, UniversityofSheffield, 1996.
[11]ParasuramanR,SheridanTB.A ModelforTypes and LevelsofHuman InteractionWithAutomation[C],IEEETransactionsonSystems,Man, andCybernetics—PartA:SystemsandHumans, 2000, 30(3):286 -297.
[12]DeardenA, HarrisonM, WrightP. Allocationof Function:Scenarios, ContextandtheEconomicsofEffort[J],InternationalJournalofHuman-ComputerStudies,2000, 52(2):289 -318.
[13] Zhang Wei, Li Daochun, Song Bifeng. Simulation of dynamic allocation of human and machine functions [J]. Ergonomics, 2005, 11 (1): 5 -7.
[14] Zhou Qianxiang, Zhou Shihua. A Model for Manned Spacecraft Man-Machine Function Assignment [J]. Ergonomics, 2003, 9(2):3-6.
[15] Zhou Shihua, Zhou Qianxiang, Qu Zhansheng. Research progress of human-machine function allocation method for complex systems [J]. Proceedings , 2004:955-958.