Abstract
An intelligent manufacturing environment employs internet-based communication and monitoring technologies for fault detection, diagnosis, and monitoring of industrial machines. The monitoring and fault detection are performed remotely without human intervention that predicts faults and ensures specific operational control. This article introduces a rational fault diagnosis process (RFDP) best suited for remote fault detection and diagnosis of CNC machine tools. The proposed process monitors different operational segments of the machine and extracts related data to validate its performance. The interconnection between the segments and fault impact are identified using the transfer learning process. The previously identified faults are used in the state training process to improve detection and diagnosis accuracy. Depending on the operational control continuity, the performance is assessed post the fault diagnosis. The learning paradigm is trained using the machine’s efficiency and rational data processing to predict the transfer states’ faults. The transfer states are modulated based on the efficiency and minimum-maximum control recommended for the CNC machine. This process’s performance is validated using detection accuracy, diagnosis recommendation, downtime, data processing rate, and processing time. From the experimental analysis, it is seen that for the varying data extraction rates, the proposed process improves detection accuracy by 10.14%, diagnosis recommendation by 8.58% and data processing rate by 7.95%, reducing the downtime by 8.85%, and processing by 11.24%.
1. Introduction
A computer numerical control (CNC) machine is an automating controlled machine that performs particular tasks using software embedded tools and techniques. CNC machines are cost effective when compared with manual machines. CNC machines mostly use microcomputers attached to machines’ plastic parts [1]. CNC is widely used to monitor and control the movements of machines. CNC improves the effectiveness and efficiency of automated companies and organizations. Moving and controlling machine axes is accomplished through the use of a CNC controller and a collection of motors and drives. The cutter’s speed and position are continually monitored and adjusted by a sophisticated feedback system on industrial equipment. CNC machines reduce the production management system’s time and energy consumption rate [2]. Fault detection in CNC machines is a complicated task to perform. Fault detection plays a major role in ensuring products’ and employees’ safety and security. Fault detection using a remote monitoring system is mostly used for CNC machines [3]. A remote monitoring system identifies the defects and faults presented in a database using a certain set of tools and functions. The remote monitoring system provides an accurate and actual set of information related to CNC machines for the fault detection process [4]. CNC machine tools are the primary production equipment in discrete manufacturing enterprises, and collecting and monitoring data is a crucial aspect of intelligent manufacturing workshops. Efficiency gains in production and a reduction in information overload are two of this technology’s most notable advantages. Data collection and monitoring of CNC machine tools in the production workshop of enterprises is the focus of this paper. Verification and classification methods are used here to calculate the actual values for further process. A remote monitoring system improves the accuracy rate of the CNC machine’s fault detection process [5].
Remote diagnostics is a procedure that uses a remote control system to identify CNC machine problems. Controlling the system’s electrical performance with remote diagnostics is a common use. The procedure of remote diagnostics recognizes CNC machines’ precise traits and patterns [6]. Remote diagnosis provides an appropriate set of data for various functions and processes in an electrical control system. Remote diagnosis even discovers complex components and information presented in CNC machines [7]. The remote diagnosis process is an automatic technique to control CNC machines’ electrical defects. Signal extraction technology is sued here to identify the electrical signals shared in CNC machines [8]. Signal extraction technology detects weak and strong signals in the automation and computation process. To identify weak signals, a technique known as weak signal identification must first examine the principles governing noise generation and the properties and correlations among signals. It is possible to identify the weak signals obscured by disturbances using applicable electronics, physics, information, and computational approaches. The signal extraction method reduces the latency rate in the identification process, which improves the efficiency level of the system [9]. Remote diagnosis also identifies the faults and problems that are occurred due to electrical control. Remote diagnosis provides an appropriate set of information for various functions in CNC machines. Remote diagnosis also produces feasible content for electrical control [10].
An intelligent technology application is an artificial intelligence (AI) embedded and analysis-related application. An intelligent technology application is an artificial intelligence (AI) embedded and analysis-related application. The intelligent application configures users’ feedback and advice via the interaction process [11]. Interaction and communication processes play a major role in intelligence applications that provide the necessary set of information for the analysis process. Intelligent application is also used for the CNC machine diagnosis process [12]. CNC machines perform AI and provide necessary features and functionalities for intelligent applications to perform certain tasks for users. Intelligent application is widely used for the CNC machine diagnosis process to reduce fault detection and prediction latency. CNC machines perform AI and provide necessary features and functionalities for intelligent applications to perform certain tasks for the users. Intelligent application is widely used for the CNC machine diagnosis process to reduce latency rate in fault detection and prediction process [13]. It is also important to guarantee effective quality control of CNC components to eliminate faulty products, limit risks, assure dimensional correctness and quality, preserve resources, decrease costs, and enhance productivity. It is good for both producers and buyers. Intelligent application is mainly used to find out or assess the system and provide necessary information for various processes. Analytical intelligence-based application is also used for the CNC machine diagnosis process [14]. Analytical intelligence identifies the important set of features and key values that are presented in CNC machines. The diagnostic accuracy rate is increased by analytical intelligence, which increases the system’s efficacy and practicality [15].
2. Contribution
An input component’s rationality defect diagnosis entails checking the correctness of the input signal in normal operation and comparison with all other relevant information. Detection of faults is vital in processes with high financial stakes but also high levels of public safety. It is possible to prevent abnormal events by spotting process flaws early on. There are a variety of methods for locating a fault. Modulating the transfer states occurs according to the efficiency recommendations and minimum-maximum control settings for the CNC machine.
3. Related Works
Peng et al. [16] introduced a new missing data tolerant fault detection method using compressive sensing for remote condition monitoring systems. A feasible collection of data for fault detection is initially gathered through the suggested method’s first step: wind turbines. There is an electrical fault detection procedure that relies on the use of spectrum analysis here. The detection and calculation accuracy are improved, while computation time is also decreased using the new technique.
Han et al. [17] proposed a long short-term memory (LSTM)-based variational autoencoder (LSTM-VAE) for the fault detection process. The proposed LSTM-VAE is mainly used for machine-related applications. The feature extraction method is implemented here to identify machines’ important features and patterns. The suggested LSTM-VAE technique delivers a high accuracy rate in the fault detection process, which improves the system’s performance and scalability.
Liu et al. [18] introduced a new intelligent monitoring system for smart factories and companies. Computer numerical control (CNC) machine tools are used here for a monitoring system. A data analysis method is used in CNC to discover the datasets required for the monitoring process. Improved flexibility and reliability can be achieved by reducing the computational complexity of factories.
Al-Naggar et al. [19] proposed an Internet of Things (IoT)-based condition monitoring method for CNC machines. The Internet of Things (IoT) is being utilized to increase the communication and interaction process’ accuracy. Predictive miniatures identification is the primary use of the new approach. An ideal collection of data is provided by identifying CNC machines’ circumstances and frequency level. The Internet of Things (IoT) improves CNC machines’ performance and efficacy.
Zhang et al. [20] designed an acoustic defect detection method using machine learning (ML) techniques. A feature extraction algorithm is utilized for the fault identification procedure that identifies the most relevant characteristics and patterns. To speed up the identification process feature extraction approach gives an actual dataset and minimizes the time consumption rate. The suggested technique achieves a high accuracy rate in the defect identification process, lowering an application’s complexity level.
Kim et al. [21] introduced a multidevice operation monitoring system using a convolutional neural network (CNN). The CNN carries out the detection of machine flaws and defects. The defect detection procedure has to know some key characteristics: the feature extraction approach identifies these key values, features, and patterns. The suggested CNN-based technique improves fault detection accuracy.
Mahmood et al. [22] developed a new reinforcement learning system-based fault detection method for wireless sensor networks (WSN). A certain set of routing protocols and tools are used here to find out the extracted location of faults presented in a machine. The method reduces the computation process’ overall latency and energy consumption rates. The proposed method reduces the computation process’ overall latency and energy consumption rates. The proposed method enhances the efficiency of the system by increasing the accuracy rate in the fault detection process.
Shicong et al. [23] introduced a remote monitoring and management system for computer numerical control (CNC) machines. Internet of Things (IoT) is used here that provide appropriate tools and methods to identify faults in CNC machines. IoT identifies the exact faults and defects that are presented in the CNC machines. Wireless sensor nodes are used here to locate the defects that reduce the energy consumption rate in the identification process. The proposed method increases the communication and interaction services rate for the users.
Koujok et al. [24] designed a machine learning (ML)-based fault diagnosis process. Diagnostic data are provided via a multiagent methodology that identifies problems and generates an appropriate collection of data for the next step in the procedure. Classifying errors based on a certain set of patterns and functions is done by using ML. Finding the root cause of the problems and generating diagnostic data are accomplished using a semisupervised approach. The suggested solution enhances the system’s efficiency by increasing the accuracy of the fault detection process.
Yun et al. [25] proposed a new weak fault detection method using the stochastic resonance (SR) approach. SR identifies the fault based on noise signals and locates the extract location of the faults. The time-delayed characteristics essential for the defect detection procedure are initially discovered. Time-delayed feedback includes what users are complaining about. Fault detection may be improved by using the suggested approach, which has a high degree of accuracy.
Jinfei et al. [26] developed a new Bayesian optimization approach-based fault tree analysis (FTA) system. The proposed method is mainly used for computer numerical control (CNC) machines. Factory CNC machines are widely used to perform tasks with a minimum energy consumption rate. FTA finds both heavy and weak faults that are presented in CNC machines. The proposed method improves the efficiency and accuracy rate of the fault diagnosis process.
Mourtzis et al. [27] proposed an Internet of Things (IoT)-based fault monitoring system for wireless sensor networks (WSN). IoT utilizes the functionalities and operations to identify a machine’s exact faults and defects. An intelligent algorithm is used here to verify the faults in predictive maintenance. A remote monitoring system provides the necessary dataset for the fault detection process, enhancing the system’s effectiveness and flexibility rate.
Chen [28] introduced a new line data monitoring system for the Internet of Things (IoT) platform. An intelligent manufacturing technique is used here to find the exact faults and defects presented in an application. Wireless sensors and nodes are used here to find both weak and heavy faults. The proposed method achieves a high accuracy rate in the fault detection process that reduces the time consumption rate in the computation process.
Zhang et al. [29] designed an industrial artificial intelligence (I-AI) approach for smart companies and factories. The proposed method is a reference model mainly used for the classification, identification, and verification process. I-AI reduces the complexity and latency rate in the computation process. The proposed method improves the effectiveness and efficiency of industries based on I-AI.
4. Proposed Diagnosis Process
The proposed operational control process is designed to regulate the functions of the CNC machines employed in an intelligent manufacturing environment. In electrical control of CNC machine tools provide a level of accuracy, efficiency, and consistency, that would be impossible to achieve in a manual process that predicts faults and ensures specific operational control. There is still no solution to monitoring industrial machines, remote fault detection, and diagnosis of modern high-precision CNC machine tools. In this article, the CNC machine tool faults prediction detection has focused on establishing the interconnection between the segments, and fault impacts are identified. Still, no outputs have represented themselves well in the industry. Since there have been no new technological breakthroughs in the experimental studies on fault detection, diagnosis and monitoring of industrial machines are commonly used. There is relatively less downtime and processing rate of CNC machine tools. The monitoring and fault detection are identified remotely without human intervention and are analyzed and controlled through monitoring sensors. An intelligent manufacturing environment’s operational data are collected and processed by a mix of software and hardware sensors. Sensors equipped with sensing units collect operational data, such as defect detection, detection, and diagnostics. These operational data are used to reduce the functions and controls of the CNC machine tools by analyzing data. The sensors’ sensing components may gather operating data such as defect detection and diagnostics. These operational data are used to reduce the functions and controls of the CNC machine tools. In the suggested RFDP, the diagnostic advice of CNC machines is used to create accurate monitoring and defect detection. Building the monitoring and fault detection process is the first step in diagnosing the faults in CNC machine tools. This mechanism causing the CNC machine tool deformations is so difficult that it is impractical to compute an analytical expression as the fault detection by use of all initial conditions and operating conditions for a CNC machine tool. The remote fault detection processes are all required to identify the best segmenting connections between the operational continuity and control changes. For example, intelligent manufacturing and remote diagnosis map the operational segments and extract related data to the fault detection to predict the fault detection, and a successive transfer learning process is used to solve the efficiency of rational fault detection. For example, intelligent manufacturing and remote diagnosis map the operational segments and extract related data to the fault detection to predict the fault detection, and a successive transfer learning process is used to solve the efficiency of rational fault detection. Figure 1 presents the proposed process.
An intelligent manufacturing environment includes a centralized state unit that controls the operations and diagnosis of CNC machines through monitoring sensors and data processing. Based on the transfer learning process, if the operational segments and fault impacts occur in CNC machines are identified. If it occurs, the state training process compared the impacts with previously identified faults for augmenting detection and fault diagnosis relies on the operational control continuity analysis observed post the fault diagnosis. Therefore, the transfer learning process is trained using rational data processing and machine efficiency in detecting faults between the transfer states. The operations of the CNC machines are trained using a transfer learning paradigm. The RFDP model operates between CNC machines and transfer states. The monitoring of CNC machines is controlled by the sensors, where intelligent manufacturing and control recommendations are made. The intelligent manufacturing for regulating the ON/OFF of the CNC machines is monitored using the inputs from the monitoring sensors . The operational data inputs from the are processed by the transfer states and then appropriate operational controls are posted. Therefore, RFDP is processed into 2 segments: operational data assessment and fault diagnosis.
4.1. Operational Data Evaluation
The monitoring sensors are in charge of gathering data from the CNC machine tools to be used in their monitoring systems. The operational data can be related to fault detection, diagnosis, and monitoring of industrial machines. In a sensing instance, the operational data observed is computed as in equation where is the active monitoring sensor in CNC machines and , , and are the minimum and maximum monitoring sensor data of industrial machines observed at different intervals. The variable and are used to illustrate the fault diagnosis and previously identified fault data. The fault diagnosis is computed as the number of fault detection identified at different operational data observations. Identifying the source of an out-of-control situation is called fault diagnosis. Diagnosing a process issue requires evaluating sensor data and process expertise to determine the present state of the plant. The impacts in because operational and diagnosis state issues in . Hence, these issues influence the fault detection at any sequence, for which the machine’s efficiency is computed as
In equation (2), the efficiency of CNC machines follows the extreme operational state and the diagnosis state based on data processing. Here, is a diagnosis state output, whereas is the operational state output, for which the appropriate assessment of remote fault detection is required. Based on and , the continuous diagnosis state processing of CNC machine tools is estimated as
Equation (3) estimates the diagnosis state processing data continuously until is active in transferring operational data from the CNC machines. In CNC machining, a solid block of material such as brass, copper, or steel is used. By using machines with numerical control, it can produce exact and precise components to a very high degree. CNC equipment often finds tools such as lathes, mills, routers, and grinders. The operational data analysis for efficiency and fault diagnosis is presented in Figure 2.
The from different cycles observed for and . This is used for correlation with the previous data, i.e., , such that max or min is classified. The max is used for efficiency analysis, whereas the min is used for continuity check. The failing validations are considered faults and are saved as that represents the interval data (Figure 2). The active operational data handling of the states depends on the downtime until which the CNC machine requires a process (such as monitoring/detection/diagnosis). The continuous data processing of the diagnosis state is analyzed using a transfer learning process. In the factory CNC machine scenario, the detained operational data are to be converted into controls through segments and fault impact analysis that must be distributed in appropriate operational continuity to improve the periodic working of all the CNC machines. In the factory CNC machine scenario, the detained operational data is to be converted into controls through segments and fault impact analysis that must be distributed in appropriate operational continuity to improve the periodic working of all the CNC machines. Open-loop or closed-loop systems regulate the CNC machine’s location throughout production. The communication between the CNC controller and the motor is one-way. Control distribution must be immediate to meet the machine’s operating segments and collect data connected to it to validate its performance. Therefore, a less connected transfer learning paradigm based on the state training process utilized for identification. The transfer learning output is used to compute the data faults and nonperiodic control distribution sequence which is obtained via identification and -based state training process. The first instance of this transfer learning process is to sample the fault detection sequence, if is detected. The concurrence of attaining machines’ efficiency and rational data processing based on at previously identified faults, computation is the beneficial solution for separating the nonperiodic sequences. The two successive processes of at diverse time intervals, say and are fed as the input to the transfer learning. From the false detection computation, instances are modelled as per equations.
Such that,
Transfer learning process estimation starts from the successive observed instances of the first state training set as . This remote false detection is the nondiagnosis occurrence; convergence is achieved if detected later in any other data processing. Hence, in the transfer learning process, the consecutive instance of is observed for unleashing the fault diagnosis in CNC machine monitoring sensor instances. The learning process is illustrated in Figure 3.
The and inputs are used in states 1 and 2 along with the mediates. The requires a change of state, i.e., or . Depending on the , the is validated for control recommendation. The complete false diagnosis recommendation results in identified for which min is required. Therefore, state 1 is repeated until the is reached. This is validated using the and recommendations provided (refer to Figure 3). The transfer learning consists of three stages viz. segments , operational continuity , and controls followed by the output. The machine’s efficiency computation and its associated faults prediction preserved by the transfer states are represented using the following equation: where is the output of the segment layer, is the operational continuity performance is assessed, and is the control recommended for CNC machines based on and , with . Equation (6) validates and is the operational control continuity for later assessing of fault diagnosis that is used for satisfying this condition. As mentioned in the above equation (6), the data processing satisfies both and for all the state training process and based conditions. The positive and negative symbols of the control recommendation produce the machine’s efficiency is to satisfy the above state training process. As per the remote fault detection of and the fault detection of and jointly produces the results of at its least possible convergence. In this proposed model, the transfer learning for the state training process follows and followed by the transfer states-based training through and . As represented in the first and consecutive instances of input for data processing, the segments and faults impact identified using the state training for processing. From the first instance, the rational data processing is as defined in the above equation (6), and hence is the output of the other operational control process, and therefore monitoring sensor output in any are retained without fault prediction. Instead, the condition for continuous operational controls is varied wherein the different operating segments, such as and impacts the state training process. In particular, the occurrence of the fault detection is either process this condition or . This is because the inputs and are processed to improve the machine’s efficiency and rational data processing such that the probability of fault occurrence in an instance is analyzed. Based on this instance, the controls recommended for detecting faults based on the conditions, i.e., or is assessed. Therefore, the recommended for the CNC machine and its associated validations of controls to are given as in the following equation:
The variables and denotes the association of segments and fault impacts detection based on the operation control. It is to be addressed that not all recommended controls can be associated with both segments and fault impacts. The control recommendation based on efficiency assessment is illustrated in Figure 4.
The different state outputs are used for different allocations to retain the machine’s efficiency. The efficiency in the existing is retained using and variants . Contrarily, and are required for modifying the existing to achieve better efficiency. Electrical connections, conductors, and switches may be tested in the continuity test mode of a digital multimeter. The continuity of a good fuse, for example, is an indication of its quality and performance. DMMs have an audible reaction when they detect that a full route has been established. The amount of a receiving water’s loading capacity can be attributable to natural background pollution or one of its current or prospective nonpoint sources of pollution. In particular, selection (due to failure) is performed for improving new . This is required for in the consecutive cycles (refer to Figure 4). Now, the machine’s efficiency, as in equation (5), is verified using the condition and for the assessments. This necessitates accurate replacements of and in equation (6) for computing the instance of machine efficiency. Substituting and in equation (6).
In equation (8), denotes the beginning of the first state, and the negative result specifies the end of the state. In equation (8), the occurrence of operational controls where relies on fault occurrence in the CNC machine, whereas but , and for control recommendation, the CNC machine must not detect any faults from the controller unit. In the state training process, and but operation data sequences are processed periodically, whereas faults are diagnosed based on the fault detection in CNC machines. The periodic working of the CNC machine helps to retain the efficiency, detection accuracy, and diagnosis recommendation of the training state instances.
5. Discussion
The performance of the proposed process is analyzed using the [30] dataset. This dataset provides the fault information of a hydraulic test rig machine. The sensor pressure, volume flow, vibration, cooling power, motor power, and temperature are located at different segments. This dataset provides the fault information of a hydraulic test rig machine. A formal representation of the above sensors in the machine is portrayed in Figure 5.
The above representation shows up different sensors located at different machine segments. From the above, the cooler, pressure, and motor power are considered for fault diagnosis evaluation. The sensor observations are performed for consistent load cycles of about 60 s. The sensor sampling rate is either 1 Hz or 100 Hz depending on its operation. The validating instances and the operating range for the cooler, pressure, and motor power sensors are tabulated in Table 1.
The above tabulation specifications are represented for a single load cycle. The continuity is determined as motor power to pressure to cooler; this sequence is a single segment for the operating machine. The range exceeding the minimum and maximum values above indicates the operation series. The mass flow rate is constant if the quantity of fluid flowing past one place is equal to the amount of fluid flowing past another. This is what the continuity equation says must happen in a steady flow. Basically, it is just a reiteration of the universal rule of mass conservation. This marks the efficient operation of the machines for when there is no downtime. Contrarily, if the above range is violated at any running cycle, then the downtime is estimated. The downtime is computed between two successful cycles; the transfer learning identifies this as a diagnosis state from which further recommendations are provided. The following illustration presents the continuity between the above sensors in identifying the downtime for 60 mins operational cycle.
The above representation of Figure 6 shows the efficiency based on the considered three sensors (alone); it is assumed that the other sensors are flawless in the considered time. The red indicates the fault identification; the motor power fault depreciates the other two sensors (due to continuity segments). If the last (cooler) is a flaw, then the diagnosis is recommended for this sensor alone. The efficiency is estimated based on the above consideration for multiple instances such that the downtime varies. The efficiency analysis for the varying modified/recommended controls is presented in Table 2.
The downtime is inconsistent for the actual control instances, as it relies on continuity and sensor faults. The modified controls are high if the downtime is high such that the operational continuity is retained through any trial. As the modification is required to be high, the segments that require controls are high that are identified from the exceeding instances as tabulated in Table 1. The efficiency varies inversely with the downtime, which is eradicated using modified controls (refer to Table 2).
5.1. Comparative Study
The comparative study for the metrics detection accuracy, diagnosis recommendation, downtime, data processing rate, and the processing time is discussed in this section. In this comparative study, the machine and data extraction rates are varied. The methods MA-FD [24], IM-IIoT [28], and MDTFD [16] are used in the comparative analysis.
5.2. Detection Accuracy
This operational data are observed through monitoring sensors and it refers to the electrical control of intelligent manufacturing CNC machine tools based on remote diagnosis and detection technology. The CNC machine achieves the high detection accuracy required for identifying faults determination at different instances using operational and diagnosis states (refer to Figure 7). The monitoring and faults detection is mitigated based on the different operational segments and extracts related data for its validations. Segmental monitoring is critical for the company, its investors, and its stakeholders in the following ways: when it comes to units and their profitability, it gives investors all they need to know. The state training process depends on previously identified faults and data processing outputs with the current fault detection. The continuous data processing is based on machines’ efficiency and the identification of the previous fault with current operational data analysis. The transfer states are changes based on control recommended for their operational data. Based on the continuous data processing, the operating segments are used for predicting faults based on the condition with detection accuracy. Therefore, the faults detection accuracy is high in CNC machines.
(a)
(b)
5.3. Diagnosis Recommendation
This proposed model achieves high diagnosis recommendation for operational data based on segments and fault impacts are used for detecting faults at CNC machine operations (refer to Figure 8). The distribution of controls in the CNC machines for remote faults detection and diagnosis of CNC machine tools is mitigated according to the condition and for analysis of the states through the transfer learning process. The operational data analysis and previous fault detection information increase fault prediction based on operational control continuity. This fault detection is addressed based on the operational state and diagnosis state processing. The fault diagnosis is performed based on the state training process using the previously identified faults to reduce the training for transfer states. Therefore, is computed for improving the faults detection and diagnosis along with the downtime at different instances. Therefore, the fault detection in operational data analysis is to be processed depending on the segment, this monitoring and fault detection has to satisfy operational control to reduce the processing time. Operational data stores (ODS) are central databases that offer operational reporting with the most recent data from numerous transactional systems. For business reporting purposes, it allows businesses to consolidate data in its original format from several locations. The proposed model performs data processing to identify faults and increase the diagnosis recommendation.
(a)
(b)
5.4. Downtime
In this proposed model for the fault diagnosis process in CNC machines with operational segment and control achieves less downtime based on performing data processing compared to the other factors as represented in Figure 9. The detection accuracy increases based on fault detection, whereas the states can be trained upon machines efficiency and rational data processing using transfer learning is decreased, and then processing rate is computed for CNC machine tools. Based on the state training process, the previously identified faults are used for improving detection accuracy and diagnosis recommendation. The fault diagnosis is assessed based on fault detection, diagnosis, and monitoring of industrial machines identified and then prevented using the proposed RFDP model. Fault diagnosis involves finding a fault by its symptoms, using knowledge, and assessing test results. CNC machines are more costly than manually controlled machines, but the prices steadily decrease with time. Basic training and abilities are all that are required for a CNC machine operator to oversee many machines. This is difficult to prevent faults in CNC machine tools in different instances. In this model, centralized data processing is required for providing control recommendations for the CNC machines. Thus, the proposed model performs the three processes with machines’ efficient output for detecting faults, and the downtime is less in this operational data processing.
(a)
(b)
5.5. Data Processing Rate
In this proposed technique, the different operational segments of the machine extract related data validations based on operational data from the CNC machines as it does not perform operational community for recommended control through transfer learning. The addressing of fault detection in appropriate and accurate operational data is analyzed from the previously identified faults for downtime and state training instances at different time intervals. The faults identification and diagnosis in CNC machine tools through the transfer learning process. Based on machine efficiency output, the monitoring and fault detection are performed remotely without human intervention identifying faults as the instance of remote diagnosis and intelligent monitoring through transfer learning, preventing faults. The continuous operational data processing contains two processes, namely segment, operational community, and control are processed with an increased processing rate. Therefore, the conditions depend on the first and consecutive instance to identify faults. In this proposed model, the state training increases the processing rate and achieves fewer fault detection, as illustrated in Figure 10.
(a)
(b)
5.6. Processing Time
In this proposed model, the control recommendation is generated through transfer learning and achieves high processing time in this rational data processing for CNC machines. This process increases diagnosis recommendation with previously identified faults compared to the other factors in remote operational control of CNC machine tools (refer to Figure 11). Based on the process, it monitors different operational segments through a transfer learning process for distributing the controls based on the condition, this output is in periodic working. In this manner, the increasing detection accuracy and diagnosis recommendation through states training (as in equations (4), (5), and (6)), and then a continuous instance of data processing identifies fault prediction is required for the operational community. In this model, downtime and fault detection are identified for maximizing the processing rate of states using transfer learning. This identified fault detection increases processing time and provides diagnosis recommendations. Hence, the detection accuracy under different operational segments produces controls administered as in equations (7) and (8) with data processing. Hence, the fault detection is identified from different operational segments of the machines with a less processing rate. The above comparative study is tabulated with the findings in Tables 3 and 4 for the machines and data extraction rate.
(a)
(b)
6. Conclusion
In this article, a rational fault diagnosis process is introduced to improve the electrical control efficiency of CNC machines. The fault diagnosis and control recommendation are performed using remote monitoring and precise data extraction. First, the CNC machine operations are classified into segments for identifying operational continuity. The data observed from different segments are used for classifying maximum and minimum operational states. This classification induces transfer learning for initial and consecutive diagnoses for administering default or modified controls. For the fault detection process, the previously extracted data are used to improve the detection accuracy. The diagnosis and operational states are switched by reducing downtime and validating the controls at different intervals. The segment’s operational continuity ensures precise fault detection and diagnosis recommendation post the cycle completion. The minimum and maximum operational states are used to improve the machine’s efficiency without increasing downtime. From the experimental analysis, it is seen that for the varying data extraction rates, the proposed process improves detection accuracy by 10.14%, diagnosis recommendation by 8.58% and data processing rate by 7.95%, reducing the downtime by 8.85%, and processing by 11.24%.
Data Availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
Acknowledgments
This paper is supported by 2021 Chongqing University Innovation Research Group Project, grant number: CXQT21039.