Machine Learning-Based Probe Fault Detection System

Abstract

The present invention relates to the technical field of probe fault detection, and particularly to a probe fault detection system based on machine learning. The system includes an information database, a fault information acquisition port, a machine learning module, and a fault location port. The information database is used to store historical fault feature data and fault location feature data; the fault information acquisition port is used to collect real-time fault feature data of the probe; the machine learning module is used to learn and train a fault determination model and a fault location model to search for and locate the faults of the probe; the fault location port is used to search for the fault location of the probe with the assistance of the machine learning module.

Description

Technical Field

[0001] The present invention relates to the technical field of probe fault detection, and particularly to a probe fault detection system based on machine learning. Background Art

[0002] When a probe fails, a series of negative impacts will occur. The faulty probe may lead to inaccurate or incorrect data, which in turn affects the system's understanding and control of the environment or process; if a key probe in the system fails, the overall performance of the system will be affected, which is particularly important for applications that require real-time decision-making and response; in some applications, such as industrial control systems or medical devices, faulty probes may pose safety risks, and misleading data or missing monitoring information may trigger unexpected events or accidents. At the same time, it may lead to the distortion of experimental results, thus affecting the repeatability and accuracy of scientific research.

[0003] Currently, machine learning and deep learning algorithms can be used to detect probe faults by training models, including supervised learning, unsupervised learning, and semi-supervised learning methods, as well as adaptive learning techniques. Through intelligent means, human resources can be freed from cumbersome work, so that valuable human resources can be invested in more valuable production work, maximizing the productivity of maintenance personnel and reducing labor costs.

[0004] For example, in the patent with the publication number CN112986880A, a probe station fault detection method and its system, a probe station and its use method are disclosed. The fault detection system includes: a storage module for storing set parameter values, a data acquisition module for collecting real-time progress data values, a data processing module, and a display module; the data processing module is used to obtain the set parameter values and real-time progress data values, compare them, and obtain the specific fault points of the probe station hardware part according to the comparison result; the display module is used to display the set parameter values, real-time progress data values, and comparison results. By using this probe station fault detection system, when the probe station fails, by comparing the set parameter values of the variables of the probe station hardware part with the real-time progress data values at the time of failure, the specific fault points of the probe station hardware part can be quickly and accurately obtained, effectively shortening the equipment downtime, improving the accuracy of spare part replacement, reducing the equipment maintenance cost, and improving the equipment utilization rate.

[0005] The above patent has the following problem: in probe fault detection, only by using the method of data comparison for fault detection, it is easy to produce misjudgment and affect the accuracy of the detection result. Summary of the Invention

[0006] In view of the deficiencies of the prior art, the present invention proposes a probe fault detection system based on machine learning. By storing historical fault feature data and fault location feature data to form an information library, collecting real-time fault feature data of the probe, converting the fault feature data into fault statistical feature data, using a machine learning model to construct a fault determination model and a fault location model, searching and locating faults, and feeding back the location information to the display side for maintenance personnel to handle, it accurately reflects the probe fault location, effectively improves the accuracy and intuitiveness of probe fault detection, and has great practical significance for improving the efficiency and accuracy of fault troubleshooting.

[0007] To achieve the above object, the technical solution of the present invention is as follows:

[0008] A probe fault detection system based on machine learning, the system includes an information library, a fault information acquisition port, a machine learning module, and a fault location port. The information library is used to store historical fault feature data and fault location feature data. Among them, the historical fault feature data is represented as a sequence (x1, x2, x3,..., x n ), x n is the nth stored historical fault feature data, and the fault location feature data is represented as a sequence (X1, X2, X3,..., X n ), X n is the fault location feature data corresponding to the nth stored historical fault feature data; the fault information acquisition port is used to collect real-time fault feature data of the probe; the machine learning module is used to learn and train the fault determination model and the fault location model to search and locate the probe fault; the fault location port is used to search for the probe fault location with the assistance of the machine learning module.

[0009] A further improvement of the present invention is that the fault information acquisition port includes a fault feature data extraction module, a data transmission module, and a fault feature data processing module. The fault feature data extraction module is used to extract the probe fault feature data obtained by collection; the data transmission module is used to transmit the extracted probe fault feature data to the fault feature data processing module; the fault feature data processing module is used to convert the fault feature data into fault statistical feature data.

[0010] A further improvement of the present invention is that the machine learning module includes a fault determination model training unit and a fault location module training unit. The fault determination model training unit is used to construct a fault determination model and run a fault determination strategy; the fault location module training unit is used to construct a fault location model and run a fault location strategy.

[0011] A further improvement of the present invention lies in that the fault location port includes a fault location model extraction module, which is used to extract the fault location model in the machine learning module to obtain fault location information.

[0012] A further improvement of the present invention lies in that the conversion of the fault feature data into the fault statistical feature data includes the following specific contents: collecting various mathematical statistical values of the historical fault feature data, forming a historical fault statistical feature vector with the various mathematical statistical values, and combining all the historical fault statistical feature vectors into the historical fault statistical feature data.

[0013] A further improvement of the present invention lies in that the fault determination model training unit is used to construct a fault determination model, and the construction of the fault determination model includes the following specific contents: constructing a fault determination model with 2 layers, taking the number of all mathematical statistical values as the number of nodes in the first layer of the fault determination model, and taking the number of faults to be determined as the number of nodes in the second layer of the fault determination model.

[0014] A further improvement of the present invention lies in that the fault determination model training unit runs a fault determination strategy, and the fault determination strategy includes the following specific contents:

[0015] S1. Taking each group of historical fault statistical feature vectors in the historical fault statistical feature data as the input of the fault determination model; taking the corresponding historical fault feature data of the group of historical fault statistical feature vectors as the output, and taking the minimization of the sum of the prediction errors between the predicted value of the fault determination model and the prediction target as the prediction target, training the fault determination model until the sum of the prediction errors reaches convergence and then stopping the training.

[0016] S2. Inputting the real-time fault statistical feature data into the trained fault determination model, and calculating the historical fault feature data in the information library with the highest matching degree with the result output by the fault determination model.

[0017] A further improvement of the present invention lies in that the matching degree calculation formula in S4 is: x i is the i-th stored historical fault feature data, and y is the result output by the fault determination model.

[0018] A further improvement of the present invention lies in that the fault location module training unit runs a fault location strategy, and the fault location strategy includes the following specific steps:

[0019] S3. Taking the historical fault feature data with the highest matching degree with the result output by the fault determination model found as the input of the fault location model;

[0020] S4. Outputting the fault location feature data to locate the fault.

[0021] A further improvement of the present invention lies in that the training method of the fault location model in S3 is as follows:

[0022] S51: Extract t groups of historical fault feature data and fault location feature data from the information library;

[0023] S52: Convert each group of historical fault feature data into the form of feature vectors, use the set of t groups of feature vectors as the input of the fault location model, use the fault location feature data predicted by each group of feature vectors as the output, use the actual fault location feature data corresponding to each group of feature vectors as the prediction target, and use minimizing the sum of the prediction accuracies of all predicted fault location feature data as the training target;

[0024] S53: The calculation formula of the prediction accuracy is: where the subscript k is the number of groups of feature vectors, f is the predicted fault location feature data, F is the actual fault location feature data, and p k is the prediction accuracy between the fault location feature data predicted by the k-th group of feature vectors and the actual fault location feature data. Train the fault location model until the sum of the prediction accuracies reaches convergence and then stop training.

[0025] The technical effects of the present invention are as follows:

[0026] By storing historical fault feature data and fault location feature data to form an information library, collecting real-time fault feature data of the probe, converting the fault feature data into fault statistical feature data, using a machine learning model to construct a fault determination model and a fault location model, searching for and locating faults, and feeding back the location information to the display side for maintenance personnel to handle, it accurately reflects the probe fault location, effectively improves the accuracy and intuitiveness of probe fault detection, and has great practical significance for improving the efficiency of fault troubleshooting. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] By reading the detailed description of the non-restrictive embodiments with reference to the following drawings, other features, objects and advantages of the present invention will become more obvious:

[0028] Figure 1 is a framework schematic diagram of the probe fault detection system based on machine learning of the present invention;

[0029] Figure 2 is a schematic diagram of the fault information collection port of the probe fault detection system based on machine learning of the present invention;

[0030] Figure 3 is a schematic diagram of the machine learning module of the probe fault detection system based on machine learning of the present invention;

[0031] Figure 4 Schematic diagram of the computer-readable storage medium structure of the present invention. Detailed implementation manners

[0032] Example 1

[0033] This example proposes a probe fault detection system based on machine learning. By storing historical fault feature data and fault location feature data to form an information library, collecting real-time fault feature data of the probe, converting the fault feature data into fault statistical feature data, using a machine learning model to construct a fault determination model and a fault location model, searching for and locating faults, and feeding back the location information to the display side for maintenance personnel to process, it accurately reflects the probe fault location, effectively improves the accuracy and intuitiveness of probe fault detection, and has great practical significance for improving the efficiency of fault troubleshooting.

[0034] As Figures 1 - 3 shown, the probe fault detection system based on machine learning, the system includes an information library, a fault information collection port, a machine learning module, and a fault location port. The information library is used to store historical fault feature data and fault location feature data. Among them, the historical fault feature data is represented as a sequence (x1, x2, x3,..., x n ), x n is the nth stored historical fault feature data, and the fault location feature data is represented as a sequence (X1, X2, X3,..., X n ), X n is the fault location feature data corresponding to the nth stored historical fault feature data; the fault information collection port is used to collect real-time fault feature data of the probe; the machine learning module is used to learn and train the fault determination model and the fault location model to search for and locate the faults of the probe; the fault location port is used to search for the fault location of the probe with the assistance of the machine learning module.

[0035] In this example, the fault information collection port includes a fault feature data extraction module, a data transmission module, and a fault feature data processing module. The fault feature data extraction module is used to extract the probe fault feature data obtained by collection; the data transmission module is used to transmit the extracted probe fault feature data to the fault feature data processing module; the fault feature data processing module is used to convert the fault feature data into fault statistical feature data.

[0036] In this embodiment, the machine learning module includes a fault determination model training unit and a fault location module training unit. The fault determination model training unit is used to construct a fault determination model and run a fault determination strategy. The fault location module training unit is used to construct a fault location model and run a fault location strategy.

[0037] In this embodiment, the fault location port includes a fault location model extraction module, which is used to extract the fault location model in the machine learning module to obtain fault location information.

[0038] In this embodiment, the conversion of the fault feature data into the fault statistical feature data includes the following specific contents: collecting the mathematical statistics values of the historical fault feature data, forming the historical fault statistical feature vectors with the mathematical statistics values, and combining all the historical fault statistical feature vectors into the historical fault statistical feature data.

[0039] In this embodiment, the fault determination model training unit is used to construct a fault determination model. The construction of the fault determination model includes the following specific contents: constructing a fault determination model with 2 layers, taking the number of all the mathematical statistics values as the number of nodes in the first layer of the fault determination model, and taking the number of faults to be determined as the number of nodes in the second layer of the fault determination model.

[0040] In this embodiment, the fault determination model training unit runs a fault determination strategy, and the fault determination strategy includes the following specific contents:

[0041] S1. Taking each group of historical fault statistical feature vectors in the historical fault statistical feature data as the input of the fault determination model; taking the corresponding historical fault feature data of the group of historical fault statistical feature vectors as the output, taking the minimization of the sum of the prediction errors between the predicted value of the fault determination model and the prediction target as the prediction target, and training the fault determination model until the sum of the prediction errors reaches convergence and then stopping the training.

[0042] S2. Inputting the real-time fault statistical feature data into the trained fault determination model, and calculating the historical fault feature data with the highest matching degree with the result output by the fault determination model in the information library.

[0043] In this embodiment, the matching degree calculation formula in S4 is: x i is the i-th stored historical fault feature data, and y is the result output by the fault determination model.

[0044] It should be noted here that this embodiment is only used to find the historical fault feature data with the highest matching degree with the result output by the fault determination model.

[0045] Embodiment 2

[0046] As shown Figures 1 - 3 in the figure, for the probe fault detection system based on machine learning, on the basis of Embodiment 1, add: the fault location module training unit runs a fault location strategy, and the fault location strategy includes the following specific steps:

[0047] S3. Use the historical fault feature data with the highest matching degree to the result output by the fault determination model found as the input of the fault location model;

[0048] S4. Output the fault location feature data to locate the fault.

[0049] In this embodiment, the training method of the fault location model in S3 is as follows:

[0050] S51: Extract t groups of historical fault feature data and fault location feature data from the information library;

[0051] S52: Convert each group of historical fault feature data into the form of a feature vector, use the set of t groups of feature vectors as the input of the fault location model, the fault location model uses the fault location feature data predicted by each group of feature vectors as the output, uses the actual fault location feature data corresponding to each group of feature vectors as the prediction target, and uses minimizing the sum of the prediction accuracies of all predicted fault location feature data as the training target;

[0052] S53: The calculation formula of the prediction accuracy is: where the subscript k is the number of groups of feature vectors, f is the predicted fault location feature data, F is the actual fault location feature data, and p k is the prediction accuracy between the fault location feature data predicted by the kth group of feature vectors and the actual fault location feature data. Train the fault location model until the sum of the prediction accuracies reaches convergence and then stop training.

[0053] It should be noted here that by storing historical fault feature data and fault location feature data to form an information library, by collecting real-time fault feature data of the probe, converting the fault feature data into fault statistical feature data, using a machine learning model to construct a fault determination model and a fault location model, searching for and locating faults, and feeding back the location information to the display side for maintenance personnel to process, it accurately reflects the probe fault location, effectively improves the accuracy and intuitiveness of probe fault detection, and has great practical significance for improving the fault troubleshooting efficiency.

[0054] It should be understood that in various embodiments of the present application, the magnitudes of the serial numbers of the above processes do not imply the order of execution, and the order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0055] It should be understood that determining B according to A does not mean determining B only according to A, but also B can be determined according to A and / or other information.

[0056] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions according to the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired network or / and a wireless network. The computer-readable storage medium can be any available medium that the computer can access, or a data storage device such as a server or data center that contains one or more collections of available media. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0057] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in the present invention can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0058] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0059] In several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only one way, and in actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of devices or units can be in electrical, mechanical, or other forms.

[0060] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0061] In addition, in each embodiment of the present invention, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0062] In the description of this specification, the descriptions referring to terms such as "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic descriptions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the described specific features, structures, materials, or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0063] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the present invention to only the specific implementation manners. Obviously, according to the content of this specification, many modifications and variations can be made. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the present invention, so that those skilled in the relevant technical fields can understand and utilize the present invention well. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A probe fault detection system based on machine learning, characterized in that the system includes an information database, a fault information collection port, a machine learning module, and a fault location port. The information database is used to store historical fault feature data and fault location feature data. Among them, the historical fault feature data is represented as a sequence (x1, x2, x3,..., xn), where xn is the nth stored historical fault feature data, and the fault location feature data is represented as a sequence (X1, X2, X3,..., Xn), and Xn is the fault location feature data corresponding to the nth stored historical fault feature data; the fault information collection port is used to collect the real-time fault feature data of the probe; the machine learning module is used to learn and train a fault determination model and a fault location model to search for and locate the faults of the probe; the fault location port is used to search for the fault location of the probe with the assistance of the machine learning module; wherein, the fault determination model includes the following specific content: construct a fault determination model with 2 layers, take the number of all statistical values as the number of nodes in the first layer of the fault determination model, and take the number of faults to be determined as the number of nodes in the second layer of the fault determination model; The fault determination strategy includes the following specific content: S1. Take each group of historical fault statistical feature vectors in the historical fault statistical feature data as the input of the fault determination model; take the historical fault feature data corresponding to this group of historical fault statistical feature vectors as the output, and use minimizing the sum of the prediction errors between the predicted value of the fault determination model and the prediction target as the prediction target, and train the fault determination model until the sum of the prediction errors reaches convergence and then stop training; S2. Input the real-time fault statistical feature data into the trained fault determination model, and calculate the historical fault feature data in the information database with the highest matching degree with the result output by the fault determination model; S3. Take the historical fault feature data with the highest matching degree found and output by the fault determination model as the input of the fault location model; S4. Output the fault location feature data to locate the fault.

2. The probe fault detection system based on machine learning according to claim 1, characterized in that the fault information collection port includes a fault feature data extraction module, a data transmission module, and a fault feature data processing module. The fault feature data extraction module is used to extract the probe fault feature data collected; the data transmission module is used to transmit the extracted probe fault feature data to the fault feature data processing module; the fault feature data processing module is used to convert the fault feature data into fault statistical feature data.

3. The probe fault detection system based on machine learning according to claim 2, characterized in that the machine learning module includes a fault determination model training unit and a fault location module training unit. The fault determination model training unit is used to construct a fault determination model and run the fault determination strategy; the fault location module training unit is used to construct a fault location model and run the fault location strategy.

4. The machine learning-based probe fault detection system according to claim 3, wherein the fault location port includes a fault location model extraction module, and the fault location model extraction module is configured to extract the fault location model in the machine learning module to obtain fault location information.

5. The machine learning-based probe fault detection system according to claim 4, wherein the conversion of the fault feature data into the fault statistical feature data includes the following specific contents: collecting various mathematical and statistical values of the historical fault feature data, forming a historical fault statistical feature vector with the various mathematical and statistical values, and combining all the historical fault statistical feature vectors into the historical fault statistical feature data.

6. The machine learning-based probe fault detection system according to claim 1, wherein The matching degree calculation formula in S4 is :; x i is the i-th historical fault feature data stored, and y is the result output by the fault determination model.

7. The machine learning-based probe fault detection system according to claim 6, wherein the training method of the fault location model in S3 is as follows: S51: Extract t groups of historical fault feature data and fault location feature data from the information library; S52: Convert each group of historical fault feature data into the form of a feature vector, use the set of t groups of feature vectors as the input of the fault location model, use the fault location feature data predicted by each group of feature vectors as the output, use the actual fault location feature data corresponding to each group of feature vectors as the prediction target, and use minimizing the sum of the prediction accuracies of all the predicted fault location feature data as the training target; S53: The calculation formula for the prediction accuracy is as follows: ; where the subscript k is the number of groups of feature vectors, f is the predicted fault location feature data, F is the actual fault location feature data, and p k is the prediction accuracy between the predicted fault location feature data and the actual fault location feature data of the k-th group of feature vectors. Train the fault location model until the sum of the prediction accuracies reaches convergence and then stop training.

Images