Mechanical equipment fault diagnosis based on wireless sensor network data fusion technology

Fang Hao; Qiuping Yang; Anjali Sharma; Vipin Balyan

doi:10.1515/pjbr-2022-0097

Article Open Access

Mechanical equipment fault diagnosis based on wireless sensor network data fusion technology

Fang Hao , Qiuping Yang , Anjali Sharma and Vipin Balyan

Published/Copyright: May 25, 2023

Published by

Become an author with De Gruyter Brill

Submit Manuscript Author Information

From the journal Paladyn, Journal of Behavioral Robotics Volume 14 Issue 1

Abstract

To save network energy consumption and prolong network life cycle in complex mechanical fault diagnosis, a research method of data fusion routing protocol algorithm based on wireless sensor network (WSN) is proposed. The specific content of the method is as follows: First, the low-energy adaptive clustering hierarchy algorithm is analyzed and discussed. On this basis, the prim route fusion algorithm is proposed to realize the effective utilization of energy and prolong the life of the network. Then, the WSN is abstracted as an undirected graph. From the perspective of saving the energy of the whole network, several current algorithms for building fusion trees are compared. The experimental results show that the prim algorithm consumes energy only after 700 rounds of clustering, while the leach clustering algorithm consumes energy only after 500 rounds. This shows that applying the prim algorithm can reduce the energy consumption of the whole network and prolong the life cycle of the network. However, the algorithm is carried out on the premise of uniform distribution of nodes, and there is a certain gap with the specific application of WSN in mechanical fault diagnosis. In the comparison of node energy consumption, it is found that compared with using the shortest path tree, using the central point of graph algorithm can greatly save the energy consumption of the node and has better performance. Practice has proved that this method can effectively remove redundant data information and solve the problem of unreliable data collected by a single sensor node. It is more suitable for the specific application of WSN in mechanical fault diagnosis.

Keywords: wireless sensors; data fusion; mechanical equipment; fault diagnosis

1 Introduction

The precision, complexity, and automation of modern mechanical equipment are increasingly demanding. Equipment failure will not only lead to huge economic losses, but also endanger personal safety. At present, the monitoring, diagnosis, and maintenance of equipment service status are mainly carried out through the monitoring and processing of equipment vibration signals. How to effectively obtain vibration signals is the key issue to be considered in the diagnosis and maintenance of mechanical equipment [1]. Nowadays, for the diagnosis and maintenance of large equipment, hundreds of mechanical vibration monitoring systems have been launched, including off-line and on-line slave machines, master-slave machines, and distributed and networked mechanical vibration monitoring systems [2]. These mechanical vibration monitoring systems have the disadvantages of high deployment cost, poor maintainability, and lack of flexibility in the process of mechanical vibration signal analysis, processing, and fault diagnosis. For some special and extreme cases, the wired sensor diagnosis mode is even more difficult to realize. Facing these problems of wired fault diagnosis system, an alternative solution is to use the emerging wireless sensor network (WSN) technology to build a wireless, distributed mechanical equipment fault diagnosis system. Among them, data aggregation is a very important technology in WSNs. In the application of complex mechanical fault diagnosis, a large number of original data collected by multiple wireless sensor nodes can be processed in the network through the data fusion algorithm to remove the redundant information and minimize the amount of data to be transmitted on the premise of meeting the fault diagnosis, so as to achieve the purpose of saving network energy consumption. Detecting the fault status of mechanical equipment through WSN data fusion technology can avoid defects in wired systems, provide the most primitive data for mechanical fault diagnosis, and improve the efficiency and success rate of system equipment diagnosis. The combination of data fusion and routing protocol is an important method to realize data processing in the network, is the key to realize data-centric routing protocol, is an important means to improve network energy efficiency, and is also the first problem to be solved by all support technologies related to data fusion [3,4] as shown in Figure 1.

Figure 1

Data fusion in WSNs.

2 Literature review

Aiming at the research of WSN fault in mechanical diagnosis, Zou et al. proposed a distributed fault detection (DFD) method, which uses the correlation of the collected quantities of adjacent nodes for fault detection [5]. Kumar and Rao proposed an event-driven WSN fault diagnosis algorithm [6]. Zhang et al. and Shen et al. proposed a fault diagnosis method using recurrent neural network [7,8]. Gong et al. proposed an improved DFD method [9]. Current research has found that a great limitation in WSNs is the low energy consumption demand of sensor nodes. Sensor nodes generally carry limited and irreplaceable energy. The energy consumption in the network mainly includes two aspects: computing and communication. The transmission of redundant data will consume too much energy to a certain extent and shorten the lifetime of the whole network. Therefore, sensor nodes should focus on local data processing and reduce long-distance data transmission, so as to reduce the huge burden of energy consumption caused by communication. To avoid the above problems, sensor networks need to use data fusion technology in the process of collecting data, delete redundant, invalid, and less reliable data, and combine the information from different nodes for fusion processing, so as to reduce the number of network data transmission. On the basis of the current research, to save network energy consumption and prolong network life cycle in complex mechanical fault diagnosis, a research method of data fusion routing protocol algorithm based on WSN is proposed. The specific content of the method is as follows: First, the low-energy adaptive clustering hierarchy (LEACH) algorithm is analyzed and discussed. On this basis, the prim route fusion algorithm is proposed to realize the effective utilization of energy and prolong the life of the network. Then, the WSN is abstracted as an undirected graph. From the perspective of saving the energy of the whole network, several current algorithms for building fusion trees are compared.

3 Method

3.1 Data fusion and routing protocol for WSNs

The data fusion technology in WSNs refers to that in the sensor network, a certain form of fusion tree is established according to certain rules. During the data transmission, the intermediate node performs fusion processing on the data, fuses the data of multiple nodes into one data that meets the user’s requirements, and only transmits the processed data to the upper node [10,11]. Although the application of data fusion technology increases the amount of calculation of intermediate nodes, it greatly reduces the redundant data in the transmission process, reduces the transmission congestion of the network, and can prolong the life cycle of the network. WSN routing protocol includes two basic functions: one is routing, which is to find the optimal path between the source node and the destination node; another function is data forwarding, that is, to correctly forward the data along the optimized path. A big limitation in WSNs is the low energy consumption demand of sensor nodes. Sensor nodes generally carry limited and irreplaceable energy. The transmission of redundant data will consume too much energy to a certain extent, shortening the lifetime of the entire network [12,13]. Therefore, sensor nodes should focus on local data processing and reduce long-distance data transmission, so as to reduce the huge burden of energy consumption caused by communication. An important implementation scheme of energy saving is to minimize the sending of redundant data, including the release phase of query requests and the return phase of sensing data to the sink node. At the same time, in monitoring applications, users generally do not need all the data collected by the network. To avoid the above problems, sensor networks need to use data fusion technology in the process of collecting data, delete redundant, invalid, and less reliable data, and combine the information from different nodes for fusion processing, so as to reduce the number of network data transmission. The combination of data fusion and routing protocol is an important method to realize the data processing in the network, the key to realize the data-centric routing protocol, an important means to improve the energy efficiency of the network, and also the first problem to be solved by all the supporting technologies related to data fusion [14].

3.2 WSN

The routing methods in WSNs can be divided into two categories according to whether data fusion is considered: address-centric (AC) routing and data-centric (DC) routing. The impact of AC routing and DC routing on energy consumption is related to the degree of data fusion, that is, the correlation between data. If there is redundancy in the original data information, DC routing can reduce the amount of forwarded data in the network, so it will show a good energy-saving effect. When there is no redundant information between the data of all data sources, the DC route cannot perform data fusion and cannot play the role of energy saving. Instead, it may consume more energy than the AC route due to the non-shortest path.

3.3 Construction of data fusion tree

Most of the data in the sensor network is transmitted from one source to the absorber node. The usual method is to use a combination of multiple holes on the surface and the suction hole to create a reverse multi-conductor melting stick, as shown in Figure 2. A. The three sensor nodes B, D, and E contain information to indicate the sink. If the transmitted data create a multicast tree recovery, the data will melt in time and reach the maximum. Tree aggregation data currently used in sensor networks typically include nearest mid-range (CNS), shortest path tree (SPT), and tree gravity [15].

Figure 2

Data fusion tree.

3.4 LEACH algorithm

LEACH is a clustering-based protocol. It is a self-organizing adaptive hierarchical routing algorithm. Because of the characteristics of WSNs, it is feasible to apply regional clustering to obtain the data monitored by the network, and it is conducive to the scalability and robustness of dynamic networks. The basic idea is to randomly select cluster head nodes in a circular manner and distribute the energy load of the whole network equally to each sensor node, so as to reduce network energy consumption and prolong network lifetime [16,17].

3.4.1 Cluster creation phase

The choice of leaching group header node depends on the number of header groups in the network and how many times each node has been the group leader to date [18]. The principle of selection is that each sensor node generates a random number between 0 and 1. If the random number is less than the first T, select the node as the leader. The calculation formula of T is shown in formula (1).

(1) T = k 1 − k × rmod N p , if n ∈ G r , 0 , if n ∉ G r ,

where K represents the number of network group leaders; N represents the total number of sensor nodes in the network; r indicates the number of completed circles.

3.4.2 Stable transmission stage

The nodes in the group collect test information and send it to the board at a given time. These nodes fall asleep at other times. This is an important way to save energy for the LEACH protocol. Stable operation time is divided into several parts. The data transmitted by the node over a given period of time occupy only a very specific part. The size of the pillars depends on the number of nodes in the cluster [19]. The time flow of one round of LEACH protocol is shown in Figure 3.

Figure 3

Time flow of one round of LEACH protocol.

After receiving the data sent from the nodes in the group, the board combines the data needed to send the mixed data to the sink. In addition, the energy consumption of the head group is very high because the head mass is far from the sink [20]. Therefore, the dynamic cluster head switching mechanism of the LEACH protocol prolongs the maximum network time. After a while, after the data are transferred, we move on to the next step.

3.4.3 Simulation experiment of LEACH protocol algorithm

In LEACH algorithm, the selection of cluster head is very important. Whether a node acts as the cluster head node can be calculated by formula (1). To make clustering proceed smoothly, the energy of nodes must be initialized, that is, the energy of 100 nodes must be randomly assigned. In LEACH algorithm, the five nodes with the largest energy are used as cluster heads and then cluster.

Through the above initialization preparations, the existing data are analyzed, and then the ID value of each node is calculated according to the distance from the node to the cluster head. Then, the algorithm is simulated and demonstrated. After clustering, the energy consumption of 100 nodes is analyzed. At 20, 40, 60, 80, and 100 nodes, the energy before and after clustering is compared, and the energy consumption of LEACH algorithm is obtained.

3.5 Prim data fusion routing algorithm

Based on the analysis and discussion of LEACH protocol, we propose a prim data fusion routing algorithm based on graph theory. Prim algorithm is the algorithm used by the minimum spanning tree in graph theory. WSNs are abstracted as an undirected connected graph, where vertices represent sensor nodes and edges represent the spacing between different sensor nodes. Now, we need to select a spanning tree to minimize the total cost, that is, the energy consumption of the entire sensor network. This problem is the problem of constructing a minimum cost spanning tree in a connected network [21,22].

3.5.1 Prim routing fusion algorithm

Figure is a binary number, G = V is a layer of voids, e is a layer of edges, and the edges are not parallel or cut into pairs of vertices. The layers of V vertices V in Figure G and edges E are represented by V(G) and E(G), respectively.

The basic contents of prim algorithm are as follows: in figure G = 〈V, E〉, where V refers to the set of edges, E refers to the set of points, and the vertex set of its spanning tree is U.

STEP 1: Put V ₀ into U;

STEP 2: Find an edge with the minimum weight among the edges ( u , v ) ∈ E of u ∈ U, v ∈ V − U and add it to the spanning tree;

STEP 3: Add the v of the edge found in STEP2 to the U set. If there are n elements in the U set, it ends. Otherwise, continue to execute STEP2. The prim algorithm is used to select the nodes, and the nodes are divided according to the energy and the distance from the node to the cluster head. From the perspective of network life cycle and energy consumption, the optimization application of prim algorithm in wireless sensor routing fusion algorithm is given in ref. [23].

The prim routing convergence algorithm is a multi-layer routing algorithm. Its design strategy is to follow the process below to continue grouping as a group layer, assuming that the nodes from layer one to layer H are selected as the header group. Take the board selection of the first set as an example, select a volunteer team by event, and send the announcement to K1 hop. The node receives a message to select the header group to be added according to the strength of the received signal; nodes that do not receive information become dashboards. Nodes that generate a random number less than the original value will be supported in the header group and data will be sent to the external header group; nodes whose random number is not less than the original value will directly identify their board or the same node based on the developer’s number and the initial price, recreate the random number, and then enter or merge the information on the board requested team. Based on the behavior, such as the LEACH algorithm, if the ball receives information from the group leader before the deadline, the time is deducted and a request to join the group is sent to the board. During the group stabilization phase, the group nodes send the data, and the board consolidates the received data and sends it to the remote station. Occasionally, an entire network is in the process of being formed and regular board elections are initiated.

3.5.2 Simulation experiment of prim data fusion routing algorithm

In the algorithm simulation, first, the energy of nodes in the network is initialized, the ID value of each node is calculated through the initialization assignment, the primary cluster head and secondary cluster head are determined according to the ID value, and then the nodes are randomly distributed. After initializing the node, we can use prim algorithm to start the current round of clustering. After clustering, we compare the energy before and after clustering. To more significantly see the superiority of this algorithm in energy, we analyze the LEACH clustering algorithm and prim clustering algorithm together and carry out multi-round simulation.

3.6 Proposal of central point of graph (CPG) algorithm

In this section, from the perspective of saving the energy of the whole network, based on the comparison of several current algorithms for constructing fusion trees, combined with the relevant knowledge of graph theory, and combined with the routing technology and data fusion technology, a CPG routing fusion algorithm with the center of graph as the fusion point is proposed. The superiority of this algorithm is proved by simulation and derivation.

3.6.1 Shortest path problem

Many optimization problems are equivalent to the shortest method of graph theory. The shortest path is usually the shortest path of the two vertical lines. The weight of the heavy road shall be equal to the weight of all the edges of the road. The shortest row is the most respectable path from each of the two rows of the vertical axis.

3.6.2 Network model

To save the total energy of the network, this article introduces a melting algorithm with a graphical representation based on the melting node. The main problem of this algorithm is how to determine the center point of the image, i.e., the joining point.

There are two cases, which are explained in Sections 3.6.2.1 and 3.6.2.2.

3.6.2.1 The determination of the center point of completely directly connected graphs

Fully directly connected graph refers to a graph in which there are direct paths between any two nodes in the graph, and the method for determining the center point is as follows:

Find the distance l 12 , l 13 . l 1 n from node l to other nodes and add them
Find L 1 . l n in sequence according to the method in step ①;
Find out min ( L x ) , then this point is the center point of the fully connected graph. Because the algorithm requires a fully direct connected graph, the model abstracted from WSNs must also be a fully direct connected graph. This requires that the communication radius and node power of WSN nodes are relatively large, which is not in line with the actual situation in the application of complex mechanical fault diagnosis. At the same time, because the fusion point has only the central node, the energy consumption burden of the central node increases. This algorithm is applicable to WSNs with relatively small sensing range. The information collected between network nodes with small sensing range also has high similarity. The maximum fusion can be carried out at the central point, and the amount of data to be transmitted by the whole network will be greatly reduced, thus reducing the network energy consumption.

3.6.2.2 The determination of the center point of non-completely direct connected graphs

A partially directly connected graph means that there is at least one path between any two nodes in the graph, but this path is not necessarily a direct path between two nodes. It may be a graph passing through some nodes in the middle. The method for determining the center point is as follows:

Repeat the Dijkstra’s algorithm n times to find the minimum distance d between each pair of vertices V i ( i = 1 , 2 , … , n ) and V j ( j = 1 , 2 , … , n ) , n is the number of nodes in the sensing area.
Measure the value of the data passing through each of these and calculate the weight of the shortest path from each vertex to the other vertex. As shown in formula (2):
(2) A = [ a ( v 1 ) , a ( v 2 ) , a ( v 3 ) , … , a ( v n ) ] = [ a 1 a 2 a 3 … a n ] S = [ S ( v 1 ) , S ( v 2 ) , S ( v 3 ) , … , S ( v n ) ] = D ⁎ A .
Judge that min { S ( S V I ) } , i is the center point.

In view of analyzing the advantages and disadvantages of the above two algorithms, we can combine the two algorithms and introduce the layered technology of WSNs. First, the sensing area of WSN is divided into several small areas. Each small sensing area is used as the first layer, and the first algorithm is used to get the center point of each sensing area; these central points are regarded as the second layer of the sensing region, and the second algorithm is applied to construct the routing fusion tree.

4 Results and discussion

4.1 Analysis of simulation results of LEACH protocol algorithm

The energy before and after clustering is compared at the places where the number of nodes is 20, 40, 60, 80, 100, etc. The comparison shows that the energy consumption of LEACH algorithm before clustering is less than that after clustering, as shown in Figure 4.

Figure 4

Comparison of energy before and after prim clustering.

4.2 Analysis of simulation results of prim data fusion routing algorithm

After initializing the node, the prim algorithm can be used to start the current round of clustering. The clustering results are shown in Figure 5.

Figure 5

Simulation diagram of prim clustering algorithm.

In the picture, we can see that the group received after the group is more hop. In this way, nodes far from the cluster head transmit data to the cluster head through the second cluster of the cluster, reducing the power consumption of the cluster head.

After clustering, we compare the energy before and after clustering and find that the energy consumed by the two is not much different from the original energy from the remaining energy of clustering. To more significantly see the energy superiority of this algorithm, we analyze the two algorithms together and conduct multi-round simulation. The simulation results are shown in Figure 6.

Figure 6

Energy comparisons between prim algorithm and LEACH algorithm.

Through the comparison between the total remaining energy and the time when they reach the cluster when the energy is exhausted, we can see that the prim algorithm only consumes the energy when the clustering reaches 700 rounds, whereas the LEACH clustering algorithm has finished the energy when the clustering reaches 500 rounds, which shows that the prim algorithm indeed prolongs the life cycle of the network. However, this algorithm is carried out under the premise of uniform node distribution, so it has certain limitations in the practical application of mechanical fault diagnosis and needs further research.

4.3 Analysis of experimental results of CPG algorithm

To measure the performance of the CPG routing fusion algorithm, it is simulated and compared to the shortest recording path SPT. Divide 30 nodes into 30 × 30 sections to create a connected image. Energy consumption is calculated according to the above energy consumption model. R is the minimum contact radius of the node (i.e., the node is not directly connected to the sink, R is the contact radius). When the node is directly connected to the sink, the contact radius is greater than R. The simulation no is shown in Table 1.

Table 1

Simulation parameters

Parameter	Numerical value
Area size	(0, 0) to (30, 30)
Number of nodes	30
Sink node location	(10, 40)
R	18
K	0.0013 pJ/bit/m⁴
Node initial energy	0.25 J

4.3.1 Simulation of SPT

In the experiment, we first initialize 30 nodes in the area of (30, 30). The coordinates of the sink node are (10, 40). In this simulation experiment, the communication radius of the node is R = 18. According to the SPT algorithm, the problem is transformed into finding the shortest path from each node in the upper node to the sink node. Use the Dijkstra algorithm to find the shortest path from each point in the figure to the sink node, as shown in Figure 7.

Figure 7

Shortest path tree.

4.3.2 Center point routing algorithm simulation

The simulation steps of the central point routing algorithm are as follows:

First, the sensing region is divided into several small regions according to the coordinate range, and the graphs composed of sensor nodes in these small regions are all fully direct connected graphs.
The center points of each graph are determined according to the algorithm of the center points of all directly connected graphs.
Each center point determined in ② is abstracted into a non-directly connected graph, and the center point of the graph is determined as the fusion node according to the algorithm of the center point of the non-directly connected graph in 3.6.2. Each node transmits data to the center point along the shortest path and then sends it to the sink node after fusion at the center point. In the SPT, because each node transmits data along the shortest path, the data fusion operation is only carried out randomly in the intermediate node, and the data cannot be fused to the greatest extent during the transmission process. In addition, the intermediate node has to wait for all the nodes that transmit data to it to complete before it can transmit data to the previous node. Therefore, there is a delay problem. The center point fusion algorithm first divides into several small sensing areas according to the sensor position. In these small areas, the similarity of the data transmitted by the nodes is relatively large, and the fusion is carried out to the greatest extent. In the second stage, the center point is determined first, and then the shortest path transmission data are carried out. In these two stages, the selection of the central point is based on saving the energy of the whole network. As shown in Figure 8, using CPG algorithm can significantly save the energy consumed by nodes.

Figure 8

Comparison of node energy consumption.

5 Conclusion

This article presents a CPG routing fusion algorithm which combines the routing technology and data fusion technology in WSNs and takes the center of the graph as the fusion node. The specific contents of this method are as follows: the data fusion technology and routing technology of WSNs are combined, the relevant knowledge of graph theory is introduced, and the prim routing fusion algorithm is proposed on the basis of LEACH algorithm, so as to achieve the effective utilization of energy and prolong the service life of the network. Then, from the point of view of saving the energy of the whole network, this article compares several current algorithms of constructing fusion tree and proposes a CPG routing fusion algorithm based on the center point of graph. By observing the energy consumption data of LEACH algorithm and prim algorithm before and after clustering, and comparing the node energy comparison diagram of SPT algorithm and center point algorithm, the effectiveness of this method is proved. Specifically: (i) Prim algorithm consumes energy only after 700 rounds of clustering, whereas the LEACH clustering algorithm has finished energy at 500 rounds, which shows that the application of prim algorithm can reduce the energy consumption of the whole network and prolong the network lifetime. However, this algorithm is carried out under the premise of uniform node distribution, and there is a certain gap with the specific application of WSN for mechanical fault diagnosis. (ii) In the comparison of node energy consumption, it is found that compared with using the SPT, using CPG algorithm can significantly save the energy consumed by nodes, and the performance is better. WSN is a new research field developed in recent years, and wireless sensor data fusion technology is an effective means to solve this problem under the numerous limitations of WSNs. It has a very broad application prospect, and its development and application will have a far-reaching impact on human life, production, and other fields.

Funding information: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Author contributions: All authors have accepted responsibility for the entire content of this manuscript and approved its submission.
Conflict of interest: The authors declare that they have no competing interests.
Informed consent: Informed consent was obtained from all individuals included in this study.
Ethical approval: The conducted research is not related to either human or animals use.
Data availability statement: The datasets and stimuli of this study are available upon reasonable request from the corresponding author.

References

[1] K. Zhang, J. Chen, T. Zhang, S. He, and Z. Zhou, “Intelligent fault diagnosis of mechanical equipment under varying working condition via iterative matching network augmented with selective signal reuse strategy,” J. Manuf. Syst., vol. 57, pp. 400–415, 2020.10.1016/j.jmsy.2020.10.007Search in Google Scholar

[2] E. Casamenti, T. Yang, P. Vlugter, and Y. Bellouard, “Vibration monitoring based on optical sensing of mechanical nonlinearities in glass suspended waveguides,” Opt. Exp., vol. 29, no. 7, pp. 10853–10862, 2021.10.1364/OE.414191Search in Google Scholar PubMed

[3] Y. Liu, T. Bao, H. Sang, and Z. Wei, “A novel method for conflict data fusion using an improved belief divergence measure in dempster–shafer evidence theory,” Math. Probl. Eng., vol. 2021, no. 2, pp. 1–15, 2021.10.1155/2021/6558843Search in Google Scholar

[4] M. S. B. Hossain, T. Rahman, N. Stojanovic, R. Rosales, T. Wettlin, S. Calabro, et al., “Transmission beyond 200 Gbit/s with IM/DD system for campus and intra-datacenter network applications,” IEEE Photonics Technol. Lett., vol. 33, no. 5, pp. 263–266, 2021.10.1109/LPT.2021.3056005Search in Google Scholar

[5] P. L. Zou, P. Wang, and C. P. Yu, “Distributed fault detection for linear time-varying multi-agent systems with relative output information,” IEEE Access, vol. 9, pp. 42933–42946, 2021.10.1109/ACCESS.2021.3066109Search in Google Scholar

[6] B. S. Kumar and P. T. Rao, “An optimal emperor penguin optimization based enhanced flower pollination algorithm in WSN for fault diagnosis and prolong network lifespan,” Wirel. Personal. Commun., vol. 127, pp. 2003–2020, 2022.10.1007/s11277-021-08765-wSearch in Google Scholar

[7] X. Zhang, R. Jiang, T. Wang, and J. Wang, “Recursive neural network for video deblurring,” IEEE Trans. Circuits Syst. Video Technol., vol. 31, no. 8, pp. 3025–3036, 2020.10.1109/TCSVT.2020.3035722Search in Google Scholar

[8] Z. Shen, Y. Bi, Y. Wang, and C. Guo, “MLP neural network-based recursive sliding mode dynamic surface control for trajectory tracking of fully actuated surface vessel subject to unknown dynamics and input saturation – science direct,” Neurocomputing, vol. 377, pp. 103–112, 2020.10.1016/j.neucom.2019.08.090Search in Google Scholar

[9] C. Gong, G. Zhu, P. Shi, and R. K. Agarwal, “Distributed fault detection and control for markov jump systems over sensor networks with round-robin protocol,” IEEE Trans. Circuits Syst. I: Regul. Pap., vol. 68, no. 8, pp. 3422–3435, 2021.10.1109/TCSI.2021.3084969Search in Google Scholar

[10] B. Yao, H. Wang, M. Shao, J. Chen, and G. Wei, “Evaluation system of smart logistics comprehensive management based on hospital data fusion technology,” J. Healthc. Eng., vol. 2022, no. 5, pp. 1–11, 2022.10.1155/2022/1490874Search in Google Scholar PubMed PubMed Central

[11] Z. Lv and H. Song, “Mobile internet of things under data physical fusion technology,” IEEE Internet Things J., vol. 7, no. 5, pp. 4616–4624, 2020.10.1109/JIOT.2019.2954588Search in Google Scholar

[12] M. Nain and N. Goyal, “Energy efficient localization through node mobility and propagation delay prediction in underwater wireless sensor network,” Wirel. Personal. Commun., vol. 122, no. 3, pp. 2667–2685, 2022.10.1007/s11277-021-09024-8Search in Google Scholar

[13] S. Manikandan and D. Kumar, “Energy efficient clustering algorithm for mobile cluster heads to enhance the lifespan of wireless sensor network,” Indian. J. Comput. Sci. Eng., vol. 12, no. 3, pp. 605–617, 2021.10.21817/indjcse/2021/v12i3/211203085Search in Google Scholar

[14] G. Singh, “A novel statistical adhoc on-demand distance vector routing protocol technique is using for preventing the mobile adhoc network from flooding attack,” Turkish J. Comput. Math. Educ. (TURCOMAT), vol. 12, no. 6, pp. 1753–1765, 2021.10.17762/turcomat.v12i6.3779Search in Google Scholar

[15] J. Bin, B. Gardiner, E. Li, and Z. Liu, “Multi-source urban data fusion for property value assessment: A case study in Philadelphia,” Neurocomputing, vol. 404, pp. 70–83, 2020.10.1016/j.neucom.2020.05.013Search in Google Scholar

[16] E. A. Abderrahmane and A. Hajraoui, “Organized selection cluster head on fuzzy low-energy adaptive clustering hierarchy protocol in three-dimensional wireless sensor networks,” Int. J. Sens. Wirel. Commun. Control, vol. 11, no. 3, pp. 362–371, 2021.10.2174/2210327910666200401154216Search in Google Scholar

[17] W. Zongshan, L. Bo, P. Yong, L. Aishan, and D. Hongwei, “LEACH algorithm of energy balance based on improved fuzzy C-means modern,” Electron. Technol., vol. 44, no. 11, p. 6, 2021.Search in Google Scholar

[18] W. Junxi and C. Guifen, “Multi hop multi-path cognitive hierarchical routing algorithm with balanced energy consumption,” Comput. Eng. Sci., vol. 43, no. 3, p. 7, 2021.Search in Google Scholar

[19] Z. Suxia, M. Hongfei, and S. Guanglu, “An energy efficient improved LEACH protocol for wireless sensor networks,” J. Harbin Univ. Technol., vol. 26, no. 3, p. 8, 2021.Search in Google Scholar

[20] W. Yachun and Z. Yan, “Dynamic optimization clustering of energy network improves,” LEACH Routing Protoc. J. Qiqihar Univ. Nat. Sci. Ed., vol. 36, no. 3, p. 7, 2020.Search in Google Scholar

[21] A. Sharma, “An optimal routing scheme for critical healthcare HTH services – an IOT perspective,” International Conference on Image Information Processing, 2017, pp. 1–5.10.1109/ICIIP.2017.8313784Search in Google Scholar

[22] J. Jayakumar, S. Chacko, and P. Ajay, “Conceptual implementation of artificial intelligent based E-mobility controller in smart city environment,” Wirel. Commun. Mob. Comput., vol. 2021, pp. 1–8, 2021.10.1155/2021/5325116Search in Google Scholar

[23] X. Liu, C. Ma, and C. Yang, “Power station flue gas desulfurization system based on automatic online monitoring platform,” J. Digital Inf. Manag., vol. 13, no. 6, pp. 480–488, 2015.Search in Google Scholar

Received: 2022-06-17

Revised: 2022-07-29

Accepted: 2022-08-08

Published Online: 2023-05-25

This work is licensed under the Creative Commons Attribution 4.0 International License.

Articles in the same Issue

https://doi.org/10.1515/pjbr-2022-0097

Keywords for this article

wireless sensors; data fusion; mechanical equipment; fault diagnosis

Creative Commons

BY 4.0