Equipment anomaly detection method and equipment anomaly detection system

Abstract

The invention discloses an equipment anomaly detection method and system, and the method comprises the steps: a local detection end generates target performance index data of to-be-detected equipment, and transmits the target performance index data to a cloud platform; the cloud platform determines target detection algorithm information corresponding to the target performance index data based on the target performance index data, and sends the target detection algorithm information to a local detection end; and based on the target detection algorithm information and the performance data of the to-be-detected device, the local detection end realizes anomaly detection of the to-be-detected device, and outputs an anomaly detection result.

Description

Technical Field

[0001] The present invention relates to the field of detection control, and more specifically to an equipment abnormality detection method and an equipment abnormality detection system. Background Art

[0002] With the widespread application of detection technology in civil and commercial fields, detection systems, especially equipment anomaly detection systems, are also facing higher requirements.

[0003] The current device anomaly detection system usually includes a local detection terminal (also called edge terminal) on the device side and a remote cloud platform (also called cloud terminal). However, the current local detection terminal will upload a large amount of collected real-time detection data to the cloud, and perform comprehensive processing and judgment on the detection data in the cloud (for example, including preprocessing, feature extraction, anomaly detection and other processes), and calculate the final anomaly detection result and send it to the local detection terminal. However, on the one hand, it is necessary to transmit a large amount of data in real time between the local detection terminal and the cloud platform via a wireless network. The data throughput per unit time is too large, resulting in the consumption of network space and resources, leading to high communication traffic (cost) and high energy consumption problems, and greatly reducing the service life of the wireless system. On the other hand, a large amount of raw data that has not been fully preprocessed is processed by complex algorithms in the cloud, which will greatly increase the amount of calculation and data processing in the cloud, and also seriously affect the timeliness of abnormal diagnosis, leading to problems such as slow system response speed and high latency.

[0004] Therefore, there is a need for an anomaly detection method and system that can achieve high real-time, high response speed, and low latency in the anomaly detection process while achieving good anomaly detection, and can effectively reduce the amount of large-scale data transmission between the cloud and the local end, reduce communication traffic and reduce energy consumption. Summary of the invention

[0005] In view of the above problems, the present invention provides a device anomaly detection method and a device anomaly detection system. The device anomaly detection method provided by the present invention can further realize a high real-time, high response speed, and low-latency anomaly detection process on the premise of achieving good anomaly detection, and can effectively reduce the large-scale data transmission between the cloud and the local end, reduce communication traffic and reduce energy consumption.

[0006] According to one aspect of the present disclosure, a device anomaly detection method is proposed, including: a local detection end generates target performance indicator data of a device to be detected, and sends the target performance indicator data to a cloud platform; the cloud platform determines target detection algorithm information corresponding to the target performance indicator data based on the target performance indicator data, and sends the target detection algorithm information to the local detection end; the local detection end implements anomaly detection of the device to be detected based on the target detection algorithm information and the performance data of the device to be detected, and outputs the anomaly detection result.

[0007] In some embodiments, the local detection end generates target performance indicator data for the device to be detected, including: the local detection end captures the performance data of the device to be detected; the local detection end determines the target performance indicator data based on the performance data based on preset rules, and sends the target performance indicator data to the cloud platform.

[0008] In some embodiments, the cloud platform determines the target anomaly detection algorithm information corresponding to the target performance indicator data based on the target performance indicator data, including: the cloud platform determines the target anomaly detection algorithm information corresponding to the target performance indicator data via a pre-trained machine learning model based on the target performance indicator data.

[0009] In some embodiments, the target anomaly detection algorithm information includes: target preprocessing model information, target feature extraction model information and target anomaly detection model information.

[0010] In some embodiments, determining the target anomaly detection algorithm information corresponding to the target performance indicator data includes: comparing the target performance indicator data with a plurality of preset target performance indicator data levels to determine the target performance indicator data level corresponding to the target performance indicator data; based on the target performance indicator data level and based on preset rules, determining the target preprocessing model information, target feature extraction model information and target anomaly detection model information corresponding to the target performance indicator data level; generating target anomaly detection algorithm information based on the target preprocessing model information, target feature extraction model information and target anomaly detection model information.

[0011] In some embodiments, the local detection end implements abnormality detection of the device to be detected based on the target detection algorithm information and the performance data of the device to be detected, including: determining the target preprocessing model, target feature extraction model and target abnormality detection model of the device to be detected based on the target detection algorithm information; preprocessing the performance data based on the target preprocessing model to obtain preprocessed performance data; extracting features of the preprocessed performance data based on the target feature extraction model to obtain device feature data; and determining the performance status of the device to be detected according to the target abnormality detection model based on the preprocessing performance data and / or the device feature data.

[0012] In some embodiments, the machine learning model is a clustering model.

[0013] In some embodiments, based on preset rules, determining the target performance indicator data based on the performance data includes: for at least one performance data among the performance data: obtaining the data type of the performance data, and obtaining the indicator gear range corresponding to the data type, wherein each indicator gear range has corresponding indicator gear information; determining the indicator gear range corresponding to the performance data, and obtaining the corresponding indicator gear information; generating the target performance indicator data based on the indicator gear information corresponding to each performance data.

[0014] In some embodiments, the equipment to be tested is a gear grinding machine, and wherein the target performance indicator data includes rotational speed data and feed rate data of the gear grinding machine; the performance data includes at least one of the rotational speed data, feed rate data, power data and vibration data of the gear grinding machine.

[0015] According to another aspect of the present disclosure, a device anomaly detection system is proposed, including a local detection terminal and a cloud platform, and the device anomaly detection system is configured to execute the method described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. The following drawings are not deliberately scaled according to the actual size, and the focus is on illustrating the main purpose of the present invention.

[0017] Figure 1A An exemplary flow chart of a device abnormality detection method according to an embodiment of the present invention is shown;

[0018] Figure 1BA schematic diagram showing a method for detecting device abnormality according to an embodiment of the present invention is shown;

[0019] Figure 2 An exemplary flow chart of a process S102 of determining target anomaly detection algorithm information according to an embodiment of the present disclosure is shown;

[0020] Figure 3 An exemplary flow chart of the process S103 of detecting anomalies of the device to be detected according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present invention.

[0022] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0023] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0024] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0025] Figure 1A An exemplary flow chart of a device abnormality detection method according to an embodiment of the present invention is shown; Figure 1B A schematic diagram of a device abnormality detection method according to an embodiment of the present invention is shown.

[0026] It should be understood that the device abnormality detection method refers to a method for detecting whether a target device (ie, the device to be detected) is in an abnormal working / performance state. It has a local detection terminal and a remote cloud platform set in the vicinity of the device to be detected.

[0027] It should be understood that the local detection end refers to a device used to implement the detection process of the device to be detected, which is arranged in the vicinity of the device to be detected to measure multiple performance data of the device to be detected during operation.

[0028] It should be understood that the local detection end may be, for example, an integrated sensor system, or may also be an integrated controller chip, etc. The embodiments of the present disclosure are not limited by the specific composition of the local detection end.

[0029] The cloud platform refers to a cloud computing platform, which is a service platform based on hardware resources and software resources to provide computing, network and storage capabilities. It may include, for example, a remote cloud server, or a remote Internet of Things system or integrated computing system. The embodiments of the present disclosure are not limited by the specific composition of the cloud platform.

[0030] The local detection terminal may communicate with the cloud platform via a wireless network, such as a cellular network, or a short-range communication network (such as zigbee, Bluetooth communication, etc.). The embodiments of the present disclosure are not limited by the specific communication method and communication protocol between the local detection terminal and the cloud platform.

[0031] Reference Figure 1A and Figure 1B The device abnormality detection method 100 includes: first, in step S101, the local detection terminal generates target performance indicator data of the device to be detected, and sends the target performance indicator data to the cloud platform.

[0032] The target performance index data is obtained based on the performance data of the device to be detected, and is intended to reflect the current overall performance level of the device to be detected. The target performance index data will be used to determine the anomaly detection algorithm corresponding to the device to be detected.

[0033] For example, the target performance indicator data can be determined based on the performance data, for example, a certain type of performance data in the performance data of the device to be detected (which can reflect the core performance or overall performance of the device to be detected) is used as the target performance indicator data, for example, the rotation speed data is used as the target performance indicator data. Alternatively, one or more types of data in the performance data can be processed to calculate the target performance indicator data.

[0034] It should be understood that, depending on the type of the device to be tested and its specific application scenario, the target performance indicator data may include, for example, multiple data types, such as rotation speed data, machine tool feed rate data, power data, vibration data, temperature data, etc. The embodiments of the present disclosure are not limited by the specific composition of the target performance indicator data.

[0035] Thereafter, in step S102, the cloud platform determines target detection algorithm information corresponding to the target performance indicator data based on the target performance indicator data, and sends the target detection algorithm information to the local detection terminal.

[0036] The target detection algorithm information refers to information used to identify an algorithm for implementing detection of the device to be detected. On the one hand, the target detection algorithm may be, for example, a single integrated algorithm or a comprehensive algorithm, or it may be multiple algorithms, for example, including a target preprocessing algorithm, a target feature extraction algorithm, and a target anomaly detection algorithm. On the other hand, the target detection algorithm information may be, for example, digital information, or text information, or a character string or binary coded information. The embodiments of the present disclosure are not limited by the type and composition of the target detection algorithm information.

[0037] It should be understood that the process of determining the target detection algorithm information corresponding to the target performance indicator data may be, for example, based on the target performance indicator data, determining the target anomaly detection algorithm information corresponding to the target performance indicator data via a pre-trained machine learning model. Alternatively, the target detection algorithm information may also be determined based on a comprehensive algorithm or a calculation function.

[0038] After the target detection algorithm information is sent to the local detection end, in step S103, the local detection end implements anomaly detection of the device to be detected based on the target detection algorithm information and the performance data of the device to be detected, and outputs anomaly detection results.

[0039] The performance data of the device to be detected refers to data that characterizes the performance of one or more aspects of the device to be detected. The performance data may be, for example, data collected by a sensor device from the device to be detected, which may be, for example, analog data, or digital data, which may be, for example, data collected in real time, or data collected in a preset cycle. The embodiments of the present disclosure are not limited by the method and type of obtaining the performance data.

[0040] The performance data may be, for example, temperature data, magnetic field data, vibration data, rotation speed data, rotation acceleration data, power data, etc. The embodiments of the present disclosure are not limited by the specific composition of the performance data.

[0041] The local detection end implements abnormality detection of the device to be detected based on the target detection algorithm information and the performance data of the device to be detected, for example including: determining one or more target algorithm models based on the acquired target detection algorithm information, and applying the one or more target algorithm models to process the performance data to obtain an abnormality detection result of the device to be detected.

[0042] It should be understood that the abnormal detection result may include, for example, an abnormal performance state or a normal performance state, or may also include a performance state and a corresponding fault point prediction and subsequent processing suggestions, etc. The embodiments of the present disclosure are not limited by the specific content of the abnormal detection result.

[0043] Based on the above, compared with the current technical solution that the local end uploads a large amount of real-time detection data to the cloud, the detection data is comprehensively processed and judged in the cloud and the results are sent to the local end. First, the present application greatly reduces the data transmission volume between the local detection end and the cloud platform end by sending the target performance index data obtained based on the performance data instead of sending a large amount of real-time collection data, so that the communication traffic cost can be reduced and the energy consumption can be reduced. Secondly, by setting the cloud platform to determine the target detection algorithm information corresponding to the target performance index data only based on the target performance index data without performing substantial complex algorithm solution, and setting the local detection end to detect the abnormality of the device to be detected based on the target detection algorithm information and the performance data of the device to be detected, and outputting the abnormal detection result, it is possible to reduce the amount of calculation, data processing and data transmission of the cloud platform, thereby ensuring the real-time and reliable data transmission between the cloud platform and the local detection end, improving the timeliness of the overall detection process, and reducing the delay problem caused by the excessive data throughput of the cloud platform, so as to take into account high real-time performance, high response speed, low delay, and low communication cost and energy consumption under the premise of achieving good abnormality detection.

[0044] In some embodiments, the local detection terminal generates target performance indicator data for the device to be detected, including: first, the local detection terminal captures the performance data of the device to be detected; second, the local detection terminal generates target performance indicator data based on the performance data based on preset rules, and sends the target performance indicator data to the cloud platform.

[0045] For example, according to actual conditions, the performance indicator data may be set to include multiple data contents, and the collected performance data may be compared with the multiple data contents, and the target performance indicator data may be determined based on the comparison result. Alternatively, the performance data collected over a period of time may be calculated and preprocessed to generate the target performance indicator data.

[0046] Next, the process of generating the target performance indicator data based on the preset rules will be described in conjunction with a specific embodiment. For example, the target performance indicator can be set as the rotational speed data of the device to be detected, and the target performance indicator data is pre-set to include two data contents: the first performance indicator data 3000cpm (turns / minute) and the second performance indicator data 6000cpm (turns / minute). The process of determining the target performance indicator data can be, for example: collecting the current rotational speed data of the device to be detected; comparing the current rotational speed data with the first performance indicator data and the second performance indicator data, and determining the performance indicator data with the smallest numerical difference with the current rotational speed data as the target performance indicator data. For example, if the current rotational speed data is 5500cpm, the second performance indicator data 6000cpm can be determined as the target performance indicator data.

[0047] Based on the above, in the present application, by setting the local detection end to generate target performance indicator data based on the performance data based on preset rules, and sending the target performance indicator data to the cloud platform, compared with directly transmitting a large amount of real-time collected data, in the present application, by first screening and preprocessing the data at the local detection end, determining the target performance indicator data that can best reflect the performance status of the device to be detected, and sending the target performance indicator data, the amount and type of data transmitted are greatly reduced, the consumption of network space and resources is reduced, and communication costs are saved.

[0048] In some embodiments, the above process of determining the target performance indicator data based on the performance data based on the preset rules can be described in more detail, for example.

[0049] First, for at least one of the performance data, the data type of the performance data is obtained, and the index gear range corresponding to the data type is obtained, wherein each index gear range has corresponding index gear information; then, the index gear range corresponding to the performance data is determined, and the corresponding index gear information is obtained. Finally, based on the corresponding index gear information of each performance data, the target performance index data is generated.

[0050] For example, if the currently acquired performance data includes speed data 4000 cpm (turns / minute) and feed rate 65%, then the index gear range corresponding to the speed data type and feed rate data type can be obtained. For example, the speed data type has a first index gear range: 2900-5900 cpm (turns / minute) and a second index gear range: 6000-8000 cpm. And the index gear information corresponding to the first index gear range is 3000 cpm, and the index gear information corresponding to the second index gear range is 6000 cpm (it should be understood that the index gear information here refers to the label information used to identify the index gear, which is intended to simply and clearly divide different gears). And the index gear corresponding to the feed rate data type includes four gear ranges: the first gear range 0-30% (the corresponding index gear information is 20%), the second gear range 31-60% (the corresponding index gear information is 50%), and the third gear range 61-100% (the corresponding index gear information is 80%).

[0051] At this time, for example, the index gear range corresponding to the performance data of the current device to be detected can be determined, and the corresponding index gear information can be obtained. Specifically, at this time, it can be known that the speed data of the current device to be detected 4000cpm (turns / minute) corresponds to the first index gear range 2900-5900cpm (the corresponding index gear information is 3000cpm), and it is known that the feed rate data of the current device to be detected 65% corresponds to the second gear range 31-60% (the corresponding index gear information is 50%), thereby obtaining the index gear information corresponding to the current two performance data as "speed: 3000cpm", "feed rate: 50%". For example, the speed = 3000cpm, feed rate = 50% can be sent to the cloud platform as the target performance indicator data.

[0052] And in some embodiments, the indicator gear information of the local detection end can, for example, correspond to the target performance indicator data gear on the cloud platform side (for example, the two can be the same data), so as to facilitate the subsequent rapid determination of the corresponding target performance indicator data gear on the cloud platform side based on the received target performance indicator data (a specific indicator gear information) and subsequent processing.

[0053] Based on the above, in this application, by setting the indicator gear range of the performance data, and setting the indicator gear information for each indicator gear range, the indicator gear information corresponding to the performance data is sent as the target performance indicator data, so that on the one hand, when processing at the local detection end, the corresponding category can be determined for each data based on the gear range, so as to better classify and assign different performance data, so that the corresponding performance range can be clarified. On the other hand, by setting the indicator gear information corresponding to the indicator gear range, the same data format and information content can be used during data interaction, which is conducive to the subsequent cloud platform to efficiently and conveniently determine the corresponding target performance indicator data gear based on the data, further streamline the data volume, and improve the efficiency of detection and processing.

[0054] In some embodiments, the cloud platform determines the target anomaly detection algorithm information corresponding to the target performance indicator data based on the target performance indicator data, including: the cloud platform determines the target anomaly detection algorithm information corresponding to the target performance indicator data via a pre-trained machine learning model based on the target performance indicator data.

[0055] It should be understood that the embodiments of the present disclosure are not limited to the specific type of the pre-trained machine learning model, which may be, for example, a clustering model, or other neural network models or comprehensive models.

[0056] It should be understood that, for example, the target performance indicator data can be input at the input end of the machine learning model, and through processing by the machine learning model (for example, through processing by its input layer, intermediate layer, fully connected layer, etc.), anomaly detection algorithm information can be obtained at the output end of the machine learning model.

[0057] Based on the above, in the present application, the target anomaly detection algorithm information corresponding to the target performance indicator data is determined through a pre-trained machine learning model based on the target performance indicator data on the cloud platform. Compared with the comprehensive calculations previously performed on the cloud platform, the cloud platform in the present application only needs to set up a single pre-trained machine learning model, and only needs to solve the target performance indicator data and determine the corresponding algorithm information, which greatly reduces the computing power of the cloud platform.

[0058] In some embodiments, the target anomaly detection algorithm information includes: target preprocessing model information, target feature extraction model information and target anomaly detection model information.

[0059] The target preprocessing model information refers to information of an algorithm model required for preprocessing the performance data of the device to be detected, such as the name, number, and identification information of the algorithm.

[0060] The target feature extraction model information refers to information of an algorithm model used to extract features from preprocessed performance data of the device to be detected, such as the name, number, and identification information of the algorithm.

[0061] The target anomaly detection model information refers to the information of the algorithm model used to judge the anomaly of the device to be detected, such as the name, number, and identification information of the algorithm.

[0062] It should be understood that according to the different data contents of the target performance indicator data of the different devices to be detected and the devices, the target preprocessing model information, target feature extraction model information and target anomaly detection model information corresponding to the devices to be detected will also be different. The embodiments of the present disclosure are not limited by the specific content of the information.

[0063] Based on the above, in the present application, by setting the target anomaly detection algorithm information including: target preprocessing model information, target feature extraction model information and target anomaly detection model information, the cloud platform can accurately and reliably determine the model type used for preprocessing, feature extraction, and target anomaly detection of the device to be detected based on the target performance indicator data of the device to be detected, thereby facilitating flexible adjustment of the corresponding anomaly detection algorithm according to the current performance of the device to be detected and its device type, thereby setting a suitable and high-precision detection algorithm for each device to be detected, thereby improving the accuracy and reliability of the detection process.

[0064] In some embodiments, the process of determining the target anomaly detection algorithm information corresponding to the target performance indicator data in step S102 may be described in more detail, for example. Figure 2 An exemplary flowchart of process S102 of determining target anomaly detection algorithm information according to an embodiment of the present disclosure is shown.

[0065] Reference Figure 2 First, in step S1021, the target performance indicator data is compared with a plurality of preset target performance indicator data levels to determine the target performance indicator data level corresponding to the target performance indicator data.

[0066] The target performance indicator data level refers to different levels of the preset target performance indicator data. The level can be, for example, a single value or a data range, and the embodiments of the present disclosure are not limited by the specific form of the data level.

[0067] For example, if the target performance indicator data includes two data types, namely, speed and feed rate, and the speed includes two cases of 3000 cpm (turns per minute) and 6000 cpm, and the feed rate includes three cases of 50%, 80%, and 100%. At this time, the target performance indicator data gear includes, for example, the following six gears: gear 1: speed 3000 cpm and feed rate 50%, gear 2: speed 3000 cpm and feed rate 80%, gear 3: speed 3000 cpm and feed rate 100%, gear 4: speed 6000 cpm and feed rate 50%, gear 5: speed 6000 cpm and feed rate 80%, gear 6: speed 6000 cpm and feed rate 100%.

[0068] Determining the target performance indicator data gear corresponding to the target performance indicator data, for example, refers to determining the target performance indicator data gear that the target performance indicator data falls into. For example, if the currently obtained target performance indicator data is: speed 3000cpm and feed rate 80%, then, for example, compared with the aforementioned six gears, the current target performance indicator data falls into gear 3, and, for example, it can be determined that the corresponding target performance data gear is gear 3.

[0069] Thereafter, in step S1022, based on the target performance indicator data level and based on preset rules, target anomaly detection algorithm information corresponding to the target performance indicator data level is determined.

[0070] It should be understood that the preset rule may be, for example, a correspondence between a data gear preset by the system and a target detection algorithm model. For example, a correspondence table between a data gear and a target detection algorithm model may be preset or pre-imported by the system. After obtaining the data gear of the device to be detected, the corresponding target detection algorithm model may be determined accordingly, thereby obtaining, for example, target preprocessing model information, target feature extraction model information, and target anomaly detection model information. For example, the target preprocessing model information, target feature extraction model information, and target anomaly detection model information may be further spliced ​​to obtain the final target anomaly detection algorithm information.

[0071] Based on the above, in the present application, when the cloud platform determines the target anomaly detection algorithm information, the target performance indicator data is compared with a plurality of preset target performance indicator data gears to determine the target performance indicator data gear corresponding to the target performance indicator data; based on the target performance indicator data gear and based on preset rules, the target anomaly detection algorithm information corresponding to the target performance indicator data gear is determined, so that the determination of the target anomaly detection algorithm information can be realized in a simple and convenient manner, thereby improving the efficiency and reliability of the target anomaly detection algorithm information generation process.

[0072] In some embodiments, the process of the local detection end realizing abnormality detection of the device to be detected based on the target detection algorithm information and the performance data of the device to be detected can be described in more detail, for example. Figure 3 An exemplary flow chart of the process S103 of detecting anomalies of the device to be detected according to an embodiment of the present disclosure is shown.

[0073] Reference Figure 3 In the process of anomaly detection, first, in step S1031, the target preprocessing model, target feature extraction model and target anomaly detection model of the device to be detected are determined based on the target detection algorithm information.

[0074] For example, the target detection algorithm information includes the name of the target preprocessing model, the name of the target feature extraction model and the name of the target anomaly detection model, and the local detection end stores various detection models. The local detection end can call the corresponding target preprocessing model, target feature extraction model and target anomaly detection model based on the name in the target detection algorithm information.

[0075] Thereafter, in step S1032, the performance data is preprocessed based on the target preprocessing model to obtain preprocessed performance data.

[0076] The preprocessed performance data refers to the data obtained after the performance data is processed by the target preprocessing model.

[0077] It should be understood that, according to the actual situation and the type of specific preprocessing model selected, the performance data can be processed in a variety of ways, such as applying a high-pass filter, a low-pass filter, or filters with different thresholds, etc., and the data can also be smoothed, denoised, and subjected to a variety of preprocessing processes according to the actual selected model. The embodiments of the present disclosure are not limited by the specific preprocessing method.

[0078] After preprocessing, in step S1033, the features of the preprocessing performance data are extracted based on the target feature extraction model to obtain device feature data.

[0079] The device feature data refers to feature data obtained after feature extraction is performed on the pre-processed performance data through a target feature extraction model.

[0080] It should be understood that, according to actual needs, the feature extraction process may include, for example, signal analysis processes in the time domain and frequency domain; and multiple data feature extraction processes such as mean, variance, peak, average, etc. The embodiments of the present disclosure are not limited by the specific content of the target feature extraction model.

[0081] After the characteristic data is obtained, in step S1034, based on the preprocessed performance data and / or the device characteristic data and according to the target anomaly detection model, the performance status of the device to be detected is determined.

[0082] It should be understood that the target anomaly detection model can, for example, perform comprehensive processing based on the preprocessing performance data and / or the device characteristic data, and output the performance status of the device to be detected. And the target anomaly detection model can, for example, be a one-class support vector machine model (oneclass svm), which, for example, only includes the preprocessing performance data and / or the device characteristic data under various normal performance states. By comparing the current preprocessing performance data and / or the device characteristic data with the data in the one-class support vector machine model, the performance status of the device to be detected can be determined simply and conveniently.

[0083] Based on the above, in the present application, the local detection end is set to determine the target preprocessing model, target feature extraction model and target anomaly detection model of the device to be detected based on the target detection algorithm information, and the performance data is preprocessed through the target preprocessing model, the features of the preprocessed performance data are extracted based on the target feature extraction model, and the performance status of the device to be detected is determined based on the preprocessing performance data and / or the device feature data according to the target anomaly detection model, so that large-scale model solution processing processes are completed by the local end. Compared with the previous setting of a large number of computing processing processes in the cloud, the detection method in the present application can greatly reduce the amount of computing in the cloud. At the same time, the best model information determined through the cloud can call the local corresponding model for processing, taking into account the reduction of cloud computing and the accuracy of detection.

[0084] In some embodiments, the machine learning model is a clustering model.

[0085] The clustering model is a machine learning model based on the clustering algorithm. Its characteristic is that it can divide the objects in the data set into different groups / clusters, and the objects in each cluster have similar characteristics. By adopting the clustering model, it is possible to simply and conveniently divide multiple different performance data levels, and realize the efficient determination of the corresponding algorithm information based on the target performance indicator data and the preset multiple target performance indicator data levels.

[0086] It should be understood that the embodiments of the present disclosure are not limited by the type of clustering model used, for example, it may be a kmeans model or other clustering models.

[0087] Based on the above, in this application, by setting the machine learning model to be a clustering model, the target performance indicator data can be divided into multiple gears within the model according to different characteristics, so that the target anomaly detection algorithm information corresponding to the target performance indicator data can be determined in a simple and convenient manner.

[0088] In some embodiments, the device to be detected is, for example, a gear grinding machine, which refers to a gear processing machine tool that uses a grinding wheel as a grinding tool to process cylindrical gears or certain gears (helical gears, bevel gears, etc.) to process tool tooth surfaces.

[0089] And wherein, the target performance indicator data includes the rotation speed data of the gear grinding machine and the feed rate data of the gear grinding machine; the performance data includes at least one of the rotation speed data of the gear grinding machine, the feed rate data of the gear grinding machine, the power data of the gear grinding machine and the vibration data of the gear grinding machine.

[0090] The rotation speed data of the gear grinding machine refers to the rotation speed data of the output shaft of the gear grinding machine.

[0091] The feed rate data of the gear grinding machine refers to the cutting speed of the gear grinding machine tool on the workpiece, that is, the speed at which the tool rotates at high speed with the main shaft and cuts forward according to a preset tool path.

[0092] The power data of the gear grinding machine refers to the overall power parameters of the gear grinding machine under the current working state, and represents the overall power consumption of the gear grinding machine.

[0093] The vibration data of the gear grinding machine refers to the vibration data of the gear grinding machine during operation.

[0094] By setting the device to be detected as, for example, a gear grinder, and setting the target performance indicator data to include the rotational speed data of the gear grinder and the feed rate data of the gear grinder; the performance data includes at least one of the rotational speed data of the gear grinder, the feed rate data of the gear grinder, the power data of the gear grinder, and the vibration data of the gear grinder, the performance status of the gear grinder during operation can be more accurately reflected, thereby improving the accuracy of the detection process.

[0095] According to another aspect of the present disclosure, a device anomaly detection system is proposed, which includes a local detection terminal and a cloud platform, and the device anomaly detection system is configured to execute the method described above and has the functions described above.

[0096] The program part of the technology can be considered as a "product" or "manufactured product" in the form of executable code and / or related data, which is participated in or realized by computer-readable media. Tangible and permanent storage media can include any memory or storage used by any computer, processor, or similar device or related module. For example, various semiconductor memories, tape drives, disk drives or any similar devices that can provide storage functions for software.

[0097] All or part of the software may sometimes communicate over a network, such as the Internet or other communications network. Such communications may load the software from one computer device or processor to another. For example: from a server or host computer of a target tracking device to a hardware platform of a computer environment, or other computer environment that implements the system, or a system with similar functions related to providing information required for target tracking. Therefore, another medium capable of transmitting software elements may also be used as a physical connection between local devices, such as light waves, radio waves, electromagnetic waves, etc., transmitted through cables, optical cables or air. Physical media used to carry carriers, such as cables, wireless connections or optical cables and the like, may also be considered as media that carry software. As used herein, unless limited to tangible "storage" media, other terms referring to computer or machine "readable media" refer to media involved in the process of executing any instructions by the processor.

[0098] The present application uses specific words to describe the embodiments of the present application. For example, "first / second embodiment", "one embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of the present application. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or multiple times in different positions in this specification does not necessarily refer to the same embodiment. In addition, some features, structures or characteristics in one or more embodiments of the present application can be appropriately combined.

[0099] In addition, it will be appreciated by those skilled in the art that various aspects of the present application may be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of the present application may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may all be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of the present application may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0100] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the invention belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0101] The above is an explanation of the present invention and should not be considered as a limitation thereof. Although several exemplary embodiments of the present invention have been described, it will be readily appreciated by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present invention. Therefore, all such modifications are intended to be included within the scope of the present invention as defined in the claims. It should be understood that the above is an explanation of the present invention and should not be considered as being limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present invention is defined by the claims and their equivalents.

Claims

1. A device anomaly detection method, comprising: The local detection terminal generates target performance indicator data of the device to be detected, and sends the target performance indicator data to the cloud platform; The cloud platform determines target detection algorithm information corresponding to the target performance indicator data based on the target performance indicator data, and sends the target detection algorithm information to the local detection terminal; The local detection end implements abnormality detection of the device to be detected based on the target detection algorithm information and the performance data of the device to be detected, and outputs the abnormality detection result.

2. The device abnormality detection method according to claim 1, wherein: The target performance indicator data of the device to be tested generated by the local detection terminal includes: The local detection terminal captures the performance data of the device to be detected; The local detection end determines target performance indicator data based on the performance data based on preset rules, and sends the target performance indicator data to the cloud platform.

3. The device abnormality detection method according to claim 1, wherein: The cloud platform determines, based on the target performance indicator data, target anomaly detection algorithm information corresponding to the target performance indicator data, including: The cloud platform determines target anomaly detection algorithm information corresponding to the target performance indicator data based on the target performance indicator data via a pre-trained machine learning model.

4. The device abnormality detection method according to claim 3, wherein: The target anomaly detection algorithm information includes: target preprocessing model information, target feature extraction model information and target anomaly detection model information.

5. The device abnormality detection method according to claim 3, wherein: Determining the target anomaly detection algorithm information corresponding to the target performance indicator data includes: Comparing the target performance indicator data with a plurality of preset target performance indicator data levels to determine a target performance indicator data level corresponding to the target performance indicator data; Based on the target performance indicator data level and a preset rule, target anomaly detection algorithm information corresponding to the target performance indicator data level is determined.

6. The device abnormality detection method according to claim 3, wherein: The local detection end implements abnormality detection of the device to be detected based on the target detection algorithm information and the performance data of the device to be detected, including: Determine a target preprocessing model, a target feature extraction model, and a target anomaly detection model of the device to be detected based on the target detection algorithm information; Preprocessing the performance data based on the target preprocessing model to obtain preprocessed performance data; Extracting features of the preprocessing performance data based on the target feature extraction model to obtain device feature data; Based on the preprocessed performance data and / or the device characteristic data, and according to the target anomaly detection model, the performance status of the device to be detected is determined.

7. The device abnormality detection method according to claim 3, wherein: The machine learning model is a clustering model.

8. The device abnormality detection method according to claim 2, wherein: Based on the preset rules, determining the target performance indicator data based on the performance data includes: For at least one of the performance data: Acquire a data type of the performance data, and obtain an indicator gear range corresponding to the data type, wherein each indicator gear range has corresponding indicator gear information; Determine the index gear range corresponding to the performance data, and obtain corresponding index gear information; Based on the indicator gear information corresponding to each performance data, the target performance indicator data is generated.

9. The device abnormality detection method according to claim 1, wherein: The equipment to be tested is a gear grinding machine. And wherein, the target performance indicator data includes the rotation speed data of the gear grinding machine and the feed rate data of the gear grinding machine; the performance data includes at least one of the rotation speed data of the gear grinding machine, the feed rate data of the gear grinding machine, the power data of the gear grinding machine and the vibration data of the gear grinding machine.

10. A device anomaly detection system, comprising a local detection terminal and a cloud platform, wherein the device anomaly detection system is configured to execute the method according to any one of claims 1 to 9.

Images