Liao Chuang 1， Li Funian 1， Yu Xingsheng 2， Yan Junfeng 2， Lin Junping 3
（1. Wuhan University of Science and Technology School of Information Science and Engineering， Hubei Wuhan 430081； 2. China Railway Fourth Survey and Design Institute Group Co., Ltd.， Hubei Wuhan 430063；3.Huazhong University of Science and Technology School of Civil and Hydraulic Engineering， Hubei Wuhan 430074）
Abstract: After the completion of bridges, the safety and stable operation of bridges depend on daily management and maintenance. The traditional manual management and maintenance methods are time-consuming and labor-intensive with poor results. Expert systems based on knowledge bases and inference engines can significantly improve the effectiveness and efficiency of bridge management and maintenance. Based on the monitoring system of the Ganjiang Grand Bridge, a set of intelligent decision expert maintenance system for bridges based on Jess was designed and developed to address real-time monitoring and manual inspection data. This system establishes a fact base and rule base of the expert system based on the current bridge management and maintenance specifications and the experience of bridge management experts, using CLIP files to describe various diseases of bridges and the corresponding maintenance measures. The system infers based on the input description of bridge diseases and automatically outputs the corresponding maintenance, repair, and treatment measures. Practice shows that this maintenance system can automatically provide optimal bridge management strategies, improving the efficiency of bridge management and maintenance and providing reliable support for bridge management and maintenance.
Keywords: Expert System; Bridge Monitoring; Bridge Maintenance; Repair Measures; Fact Base; Rule Base
Classification Number: TN02⁃34; TP393 Literature Identification Code: A
Article Number: 1004 ⁃ 373X（2022）01⁃0110 ⁃04
After the completion of bridges, they are long-term affected by rain and snow, chemical corrosion and oxidation, and the heavy pressure of passing vehicles, leading to structural damage and ultimately resulting in bridge collapse accidents, causing significant casualties and property losses. Therefore, the management and maintenance of bridges become a top priority [1⁃2]. Traditional bridge management and maintenance are supervised by inspection personnel at all times. When a bridge disease occurs, the inspection personnel report the situation to the management personnel, who analyze the diseases and propose optimal repair and treatment measures. This method consumes a lot of manpower and material resources, and some diseases may not be understood by ordinary management personnel, requiring consultation with experienced bridge management experts, thus increasing additional management costs. Moreover, the optimal time for treatment may be delayed, making the traditional bridge management method time-consuming, labor-intensive, and ineffective. Therefore, it is necessary to design an expert system with intelligent decision-making capabilities for bridge diseases. An expert system is also known as an intelligent computer system based on knowledge in a specific field, combining years of experience and corresponding professional knowledge of experts in that field to solve problems that only domain experts can address [3⁃4].
This system’s bridge management expert system will compile maintenance measures based on relevant bridge management specifications and the experience of bridge management experts, and then establish the expert system knowledge base for these maintenance measures, forming an expert system with intelligent management decision-making capabilities. When various diseases occur in the bridge, management personnel only need to describe the occurrence of the disease in the expert system, which will infer based on the input description and ultimately propose optimal maintenance, repair, and treatment measures for the bridge diseases.
1 Ganjiang Grand Bridge Monitoring and Maintenance System
The Ganjiang Grand Bridge is over 2 km long, located downstream at the confluence of the Zhangjiang and Gongjiang tributaries of the Ganjiang River. The main bridge structure is a tower-beam separation form with a semi-floating system [5]. After the bridge was completed, a relatively complete monitoring system was installed to monitor and analyze the Ganjiang Bridge in real-time, providing reliable support for the subsequent bridge management and maintenance. The monitoring of the Ganjiang Grand Bridge mainly includes the structures on the Nanchang side, the Ganzhou side, the bridge deck, and the auxiliary structures of the bridge, with a total of over 400 various sensors, including wind speed and direction, temperature and humidity, GPS, strain, deflection, acceleration, cable force, support displacement, and track monitoring, among others, as shown in Figure 1.
Figure 1 Monitoring System Structure Diagram
From the structure diagram of the monitoring system, it is clear to see the position of each sensor on the bridge. When abnormal data occurs in the sensors, the monitoring system can immediately identify which sensor on the bridge has a problem, and based on the abnormal data, the management system can automatically provide management suggestions and strategies. This ensures the timeliness of bridge maintenance and provides efficient management support for the bridge’s health and safety.
2 Expert System Based on Jess
2.1 Introduction to Expert Systems
An expert system is also a type of computer system, but it differs from ordinary computer systems in that it is a program system with specialized knowledge and practical experience in a particular field. When encountering problems, the system autonomously consults relevant problem descriptions in the knowledge base, effectively inferring the problem, ultimately solving complex issues that only domain experts can address.The structure of the expert system [6] is shown in Figure 2.
Figure 2 Structure Diagram of Expert System
2.2 Jess Expert System Shell Principle
Jess is a Java-based expert system shell developed by the Sandia National Laboratories in 1995 as an enhanced version of CLIPS. In addition to inheriting the advantages of CLIPS such as good portability and low development tool and hardware costs, Jess has many unique features, such as supporting forward and backward reasoning [7], and can directly call Java class libraries. Jess also includes interfaces developed for data interaction between Java and Jess to facilitate integration calls from Java to Jess. These advantages and features allow the expert system to combine with Java, making it very convenient to apply in different systems.Jess’s core consists of a fact base, rule base, and inference engine [8], where the rule base and fact base together form the knowledge base, as shown in Figure 3.
Figure 3 Jess Structure Diagram
2.3 Rete Matching Algorithm in Jess
The efficient forward and backward reasoning of Jess is implemented through the Rete algorithm, which is an effective solving mechanism for complex many-to-many problems. This algorithm was proposed by Charles L. Forgy of Carnegie Mellon University in 1974 [9⁃10]. Common expert systems have characteristics of structural similarity and temporal redundancy. The reason for the efficient reasoning capability of the Rete matching algorithm is that expert systems using the Rete algorithm utilize these two characteristics to reduce the number of matching operations during fact assertions, achieving efficient reasoning. When the data in the fact set changes, the system performs effective matching, and the resulting state after matching is stored in nodes [11]. When the data changes again in the fact set, since most state results remain unchanged, only a small portion changes. At this time, the Rete algorithm avoids a large number of repetitive calculations by retaining the states from the previous matching process in nodes, achieving efficient reasoning matching. However, since retaining the states from the matching process requires a large amount of memory, this matching algorithm sacrifices memory space for execution time, consuming more memory, so performance and memory trade-offs should be considered during development.
3 Establishment of the Maintenance Expert System
3.1 Establishment of the Fact Base
The fact base of this system consists of a collection of facts such as bridge disease categories, locations, and descriptions. Establishing the fact base for the management expert system means creating a storage system to store records of bridge disease categories, occurrence locations, and descriptions recorded by personnel during on-site inspections. Since different types of bridge diseases correspond to different descriptions, each disease description has its own number of parameters. For example, when a bridge experiences water leakage, the disease description includes three parameters: leakage location, leakage phenomenon, and leakage range; when a bridge has cracks, the description includes six parameters: crack location, length, width, depth, development direction, and cracking state; when a bridge experiences degradation and spalling, the description includes five parameters: degradation phenomenon, exposed reinforcement condition, degradation radius, degradation depth, and strength reduction ratio. This system establishes a working memory as shown in Figure 4 to accommodate each type of disease description with different parameter counts, including single-string attribute slots, single-string attribute slots, and multi-string attribute slots.
Figure 4 Working Memory Diagram
The categories of bridge diseases and their occurrence locations recorded by inspection personnel are stored in the first two single-string attribute slots in Figure 4, while multiple descriptive parameters of the diseases are stored in the multi-string attribute slot. This working memory is designed to accommodate all disease situations that may occur in bridges.
3.2 Establishment of the Rule Base
The establishment of the rule base for the expert management system [12] requires referencing current bridge management standards and consulting experienced bridge management experts to summarize the possible diseases of bridges and the corresponding management and maintenance measures, which are then converted into CLP files corresponding to bridge disease situations using the Jess expert system development language [13]. The rule base of the expert management system consists of multiple CLP files. In this system, these CLP files are stored in a MySQL database. When the expert system infers based on bridge diseases, it connects to the database via JDBC to retrieve the corresponding CLP files for bridge management inference. To improve inference efficiency, the descriptions of bridge disease situations and management measures are stored in two different tables in the rule base, with each management measure identified by a unique primary key ID. When the system infers the ID code of the bridge management measure based on the rules in the rule base, it finds the corresponding bridge management measure based on the ID to complete the inference for bridge disease management.The following code is a template code for the CLP file when water leakage occurs on the bridge deck.
(defrule example(template(type?tp&:(eq?tp ” water leakage “)))//Disease category is water leakage
(template(location?loc&:(eq?loc ” deck “)))//Disease location is the bridge deck
(template(num_des?nd&:(and(member>/</=drip leakage$?//Disease 1 is dripping//Disease 2 is wet//Management measures
4 Bridge Inspection Function
4.1 Formulation of Inspection Plans
To ensure the safe operation of bridges, relying solely on video monitoring is far from sufficient. This system adopts a combination of video surveillance and on-site inspections to ensure the safe operation of bridges. Bridge management personnel will periodically formulate on-site inspection plans based on the actual conditions of the bridge, including the personnel involved in the inspection, inspection time, and types of inspections. The bridge inspection plan is shown in Figure 5.
Figure 5 Bridge Inspection Plan Diagram
For example, when sensor data indicates that the bridge may have an abnormal situation, personnel will immediately formulate an on-site inspection plan for the bridge, then arrange for personnel to conduct on-site inspections and record the situation on-site to ensure timely bridge management.
4.2 Input of Inspection Results
After inspection personnel conduct on-site inspections of the bridge, they will input the inspection results based on the actual inspection situation, including the personnel involved in the inspection, inspection time, inspection items, and any diseases found during the inspection. The input of inspection results is shown in Figure 6.
Figure 6 Bridge Inspection Result Diagram
Based on the diseases found in the bridge, a preliminary judgment of the disease category is made, and then the bridge management function module interface is entered, where effective reasoning is performed based on the disease description to propose optimal bridge management and maintenance suggestions.
5 Intelligent Management Expert System Function
After a bridge disease occurs, the intelligent management expert system function module will perform reasoning matching based on the bridge data monitored by the system and the disease results obtained from on-site inspections by personnel, then provide optimal bridge management suggestions [14]. The bridge management function interface is divided into four parts: bridge disease category area, bridge disease location area, bridge disease description area, and bridge disease repair measures area. When a bridge disease occurs, the user first selects the disease category in the disease category area, then selects the location of the disease occurrence in the disease location area, and then selects the disease description in the disease description area, and finally clicks the diagnosis button on the far right. At this point, the expert system will perform reasoning matching based on the user’s input of bridge disease category, disease occurrence location, and disease description, and the inferred results will be displayed in the disease repair measures area, as shown in Figure 7.
Figure 7 Management Function Interface Diagram
Bridge management personnel will manage and maintain the bridge according to the results inferred by the management expert system to eliminate bridge diseases.
After the completion of the bridge, to ensure the safety and long-term stability of the bridge, this paper designs an intelligent computer system, namely an expert system. By referencing bridge management manuals and consulting experienced bridge management experts, various possible disease situations and solutions for bridges are compiled into rules stored in a database. When a bridge disease occurs, users only need to analyze the disease category, occurrence location, and disease situation based on the bridge’s disease and then input this into the computer system. The system will automatically reason based on the user’s input and ultimately provide optimal solutions for bridge diseases. The bridge management decision expert system module of this system has been put into practical use with good operational results, promptly proposing reasonable repair and treatment measures for bridge diseases, providing effective support for the long-term stable operation of bridge structures.
[1] Li Chenglin. Intelligent Bridge Safety Monitoring System [J]. China Science and Technology Investment, 2018(33): 41.
[2] Su Yunpeng, Sun Wenle. A Brief Discussion on the Management and Maintenance of Road Bridges [J]. Building Engineering Technology and Design, 2019(35): 1377.
[3] Song Yi, Jiao Xianlong. Design and Development of Intelligent Network Fault Management System Based on Jess and SNMP [J]. Electronic Design Engineering, 2016, 24(14): 49⁃51.
[4] Qi Fang. Research on the Application of Expert Systems in Fault Detection [J]. Communication World, 2019, 26(11): 152 ⁃ 153.
[5] Yang Yanli, Yan Aiguo. Key Technologies for the Construction Line Control of the Ganjiang Grand Bridge’s Ballastless Track [J]. China Railway, 2019(11): 104 ⁃ 108.
[6] Muhammad H H, Arshad M R H M, Azmi M D. Mobile Decision Support System with Dynamic Knowledgebase Using Decision Tree and Case-Based Reasoning [J]. International Journal of Business Information Systems, 2018, 29(1): 42 ⁃58.
[7] Chen Hongxi. Research on the Mechanism of Backward Chaining in Jess and the Development Mode of Diagnostic Expert Systems [J]. Journal of Lanzhou Petrochemical Vocational Technical College, 2015, 15(1): 32⁃ 34.
[8] Peng Hualiang, Shen Shulong, Li Jun, et al. Design of Fault Diagnosis Expert System Based on Fault Tree [J]. Control Engineering, 2019, 26(3): 584⁃588.
[9] Hao Tianyu, Li Yingshun, Shi Yang, et al. Design of an Automatic Loading System Fault Diagnosis System Based on the Rete Algorithm [J]. Electronic Design Engineering, 2018, 26(21): 38⁃41.
[10] Zhang Jingli, Zhao Chen, Li Xiaoye, et al. Comparison of Prolog and Jess Inference Algorithms [J]. Computer Programming Skills and Maintenance, 2017(15): 28⁃29.
[11] Sun Xin, Yan Ximin, Shang Yuming, et al. An Improved Rete Algorithm Based on Shared Degree Model [J]. Acta Automatica Sinica, 2017, 43(9): 1571 ⁃ 1579.
[12] Li Feng, Xia Li. Transformer Fault Monitoring Expert System Based on Rule Base [J]. Computer Science, 2016, 43(z2): 564⁃567.
[13] Wu Yi, Lü Jiajiang. Intelligent Diagnostic Expert System for CNC Machine Tools Based on CLIP Platform [J]. Science and Technology Innovation, 2019(32): 64⁃65.
[14] Liu Chuqiao, Wu Xianguo, Qin Wenwei, et al. Research and Development of Intelligent Decision Expert System for Operation and Maintenance of Rail Transit Tunnels [J]. Railway Standard Design, 2019, 63(11): 126 ⁃ 131.
Liao Chuang (1995—), male, from Tianmen, Hubei, master’s student, majoring in wireless communication and data monitoring.
Li Funian (1979—), male, from Wuhan, Hubei, doctoral student, associate professor, major research direction includes wireless sensor networks and optimization, and the application of wireless communication technology in industrial control.
Yu Xingsheng (1980—), male, from Wuhan, Hubei, senior engineer, major research direction includes bridge safety and monitoring.
Yan Junfeng (1979—), male, from Wuhan, Hubei, senior engineer, major research direction includes road and railway engineering.
Lin Junping (1996—), male, from Hubei, master’s student, major research direction includes structural dynamic analysis and damage identification.
Click the link below to view historical articles
CNKI’s continuous price increase is suspected of monopolization; how should academia and commerce balance?
Directory of Modern Electronic Technology 2019 Issue 12
In 2017, which universities published the most SCI papers and authorized patents?
Modern Electronic Technology selected in Overview of Chinese Core Journals
Academician of the Chinese Academy of Sciences: The dominance of SCI in papers stifles scientific creativity
How to write a ‘Nature’ article overnight, listen to what academicians say!
