Modeling method and device of fault diagnosis model, equipment and storage medium

Abstract

The invention discloses a fault diagnosis model modeling method and device, equipment and a storage medium, and the method comprises the steps: determining an interval frequency pair according to the value of signal data in a reference operation detection signal and an interval range, and determining a position set according to the interval range corresponding to the interval frequency pair and the signal data; determining a reference signal subset corresponding to the reference operation detection signal in the interval frequency pair according to the position interval range, and determining a non-reference signal subset corresponding to the non-reference operation detection signal according to data time corresponding to signal data in the reference signal subset; taking the reference signal subset and all the non-reference signal subsets as sub-models corresponding to sample input interval frequency pairs to obtain a model diagnosis result; and determining a model comprehensive diagnosis result according to all the model diagnosis results, determining a single accurate coefficient according to the model diagnosis result output by each sub-model, and adjusting a weight coefficient of the sub-model according to the model comprehensive diagnosis result and the single accurate coefficient of each sub-model to obtain a fault diagnosis model.

Description

Technical Field

[0001] The present invention relates to the field of fault diagnosis technology, and in particular to a fault diagnosis modeling method, apparatus, device and storage medium. Background Technology

[0002] With the widespread application of industrial robots in various fields, the demand for real-time fault diagnosis and predictive maintenance of their core components, such as rotary vector (RV) reducers, is rapidly increasing.

[0003] Currently, after using sensors for condition monitoring of industrial robots to collect various monitoring data during different working states of the industrial robots, the monitoring data is directly input into the fault diagnosis model to determine the possible fault types of the RV reducer.

[0004] However, since the signal attribute factors of the various monitoring data collected above may interact and interfere with each other in practice, such as changes in speed signals affecting current, vibration and other signals, diagnosing faults based on the above fault diagnosis model will have an adverse effect on the reliability and generalization of the diagnosis results, leading to serious deterioration of the diagnosis results. Summary of the Invention

[0005] This invention provides a modeling method, apparatus, device, and storage medium for a fault diagnosis model, which improves the robustness and diagnostic accuracy of the model and outputs more reliable results.

[0006] In a first aspect, embodiments of the present invention provide a modeling method for a fault diagnosis model, the method comprising:

[0007] Based on the numerical values ​​and interval ranges of the signal data in the baseline operation detection signal, interval frequency pairs are determined. For each interval frequency pair, the location set is determined based on the interval range and signal data corresponding to the interval frequency pair.

[0008] Based on the location interval range of the location set, determine the reference signal subset corresponding to the interval frequency pair of the reference operation detection signal, and determine the non-reference signal subset corresponding to each non-reference operation detection signal based on the data time corresponding to the signal data in the reference signal subset.

[0009] The reference signal subset and all non-reference signal subsets are used as samples and input into the sub-model corresponding to the interval frequency pair to obtain the model diagnosis results output by the sub-model.

[0010] The overall model diagnosis result is determined based on the model diagnosis results output by the corresponding sub-models according to the interval frequency. The single accuracy coefficient is determined based on the model diagnosis results output by each sub-model. The weight coefficients of the sub-models are adjusted according to the overall model diagnosis result and the single accuracy coefficients of each sub-model to obtain the modeled fault diagnosis model.

[0011] The fault diagnosis modeling method provided in this invention determines a baseline operating detection signal from multiple sources and different types of detection signals, enabling preliminary signal analysis based on one of the detection signals. Simultaneously, it divides the signal data in the baseline operating detection signal into interval ranges and determines the frequency of signal data within each interval range, thereby determining the location set corresponding to the signal data in each interval range. This provides a basis for selecting signal data from related regions in other operating detection signals. Based on the location interval range corresponding to each interval range, it determines the baseline signal subset corresponding to the interval frequency pair of the baseline operating detection signal, achieving reasonable clustering and division of the signal data in the baseline operating detection signal based on interval frequency pairs. According to the data time corresponding to the signal data in the baseline signal subset, it selects data from each non-baseline operating detection signal that matches the acquisition time point of the signal data in the baseline signal subset, forming a non-baseline signal subset. This achieves the selection of detection signals from other operating detection signals that match the acquisition time point of the baseline operating detection signal, accurately and reasonably reflecting the operating status of the RV reducer during the acquisition period, using one operating detection signal as a baseline. Using a frequency pair of intervals as a benchmark, the benchmark signal subset and non-benchmark signal subset corresponding to that frequency pair are used as samples to train the sub-model corresponding to that frequency pair. This ensures that the trained sub-model has a stronger adaptability to the signal data corresponding to that frequency pair after training, thereby improving the fault detection accuracy of the sub-model for that frequency pair. Simultaneously, by adjusting the weight coefficients of the sub-models based on the model's comprehensive diagnostic results and individual accuracy coefficients, the weight coefficient ratio of each sub-model in the fault detection model can be adjusted according to the detection data in different frequency pairs and the model's diagnostic results. This integrates the sub-models into a comprehensive fault diagnosis model, improving the model's robustness and diagnostic accuracy, and outputting more reliable results.

[0012] Secondly, embodiments of the present invention also provide a modeling apparatus for a fault diagnosis model, the apparatus comprising:

[0013] The first determining module is used to determine interval frequency pairs based on the value and interval range of the signal data in the reference operation detection signal, and for each interval frequency pair, to determine the position set based on the interval range and signal data corresponding to the interval frequency pair.

[0014] The second determining module is used to determine the reference signal subset corresponding to the reference operation detection signal in the interval frequency pair based on the position interval range of the position set, and to determine the non-reference signal subset corresponding to each non-reference operation detection signal based on the data time corresponding to the signal data in the reference signal subset.

[0015] The output module is used to take the reference signal subset and all non-reference signal subsets as samples and input them into the sub-model corresponding to the interval frequency pair to obtain the model diagnosis results output by the sub-model.

[0016] The adjustment module is used to determine the overall model diagnosis result based on the model diagnosis results output by the corresponding sub-models according to the interval frequency, determine the single accuracy coefficient based on the model diagnosis results output by each sub-model, and adjust the weight coefficient of the sub-models according to the overall model diagnosis result and the single accuracy coefficient of each sub-model to obtain the modeled fault diagnosis model.

[0017] Thirdly, embodiments of the present invention also provide an electronic device, the electronic device comprising:

[0018] At least one processor; and a memory communicatively connected to the at least one processor; wherein,

[0019] The memory stores a computer program that can be executed by at least one processor, such that the at least one processor is able to execute the modeling method of the fault diagnosis model of any embodiment of the present invention.

[0020] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing computer instructions, which are used to cause a processor to execute a modeling method for a fault diagnosis model according to any embodiment of the present invention.

[0021] Fifthly, embodiments of the present invention also provide a computer program product, including a computer program that, when executed by a processor, implements a modeling method for a fault diagnosis model according to any embodiment of the present invention.

[0022] It should be noted that the aforementioned computer instructions may be stored, in whole or in part, on a computer-readable storage medium. This computer-readable storage medium may be packaged together with the processor of the fault diagnosis modeling device, or it may be packaged separately from the processor of the fault diagnosis modeling device; this application does not impose any limitations on this.

[0023] The descriptions of the second, third, fourth, and fifth aspects in this application can be referred to the detailed description of the first aspect; and the beneficial effects of the descriptions of the second, third, fourth, and fifth aspects can be referred to the analysis of the beneficial effects of the first aspect, which will not be repeated here.

[0024] In this application, the name of the modeling device for the aforementioned fault diagnosis model does not limit the equipment or functional module itself. In actual implementation, these equipment or functional modules may appear under other names. As long as the function of each equipment or functional module is similar to that of this application, it falls within the scope of the claims of this application and its equivalents.

[0025] These or other aspects of this application will become more readily apparent in the following description. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 A flowchart illustrating a modeling method for a fault diagnosis model provided in an embodiment of the present invention;

[0028] Figure 2 A flowchart illustrating a modeling method for another fault diagnosis model provided in an embodiment of the present invention;

[0029] Figure 3 This is a schematic diagram of the structure of a modeling device for a fault diagnosis model provided in an embodiment of the present invention;

[0030] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0031] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0032] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0033] The terms “initial” and “target” in the specification and drawings of this application are used to distinguish different objects or to distinguish different treatments of the same object, rather than to describe a specific order of objects.

[0034] Furthermore, the terms "comprising" and "having," and any variations thereof, used in the description of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0035] Before discussing the exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but may also have additional steps not included in the figures. The process can correspond to a method, function, procedure, subroutine, subroutine, etc. Moreover, without conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0036] It should be noted that in the embodiments of this application, the words "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0037] In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0038] Figure 1 This is a flowchart illustrating a modeling method for a fault diagnosis model provided in an embodiment of the present invention. This embodiment is applicable to modeling a fault diagnosis model for an RV reducer. The method can be executed by a fault diagnosis modeling device, which can be implemented in hardware and / or software and can be configured in an electronic device. In this embodiment, the electronic device can be a computer or server used by maintenance personnel. (Continue referring to...) Figure 1 This embodiment specifically includes the following steps:

[0039] S101. Determine the interval frequency pair based on the value and interval range of the signal data in the reference operation detection signal. For the interval frequency pair, determine the position set based on the interval range and signal data corresponding to the interval frequency pair.

[0040] The reference operation detection signal is a pre-set monitoring data signal from the collected operating status detection data of the RV reducer, which serves as a reference. In this embodiment, the reference operation detection signal can be a speed detection signal, a current detection signal, or a torque detection signal, etc. The signal data is at least one state data point collected at preset intervals within one acquisition cycle of the operation detection signal. The interval range is a range divided according to the value range of the signal data in the reference operation detection signal. The interval frequency pair is used to characterize the total frequency of detection signal values ​​falling within an interval range. In this embodiment, the interval frequency can be represented as "interval range - frequency".

[0041] Specifically, firstly, each industrial robot is equipped with multi-source sensor hardware on its surface or within its structure to collect various operational detection signals (condition monitoring data). For example, angle sensors, current sensors, vibration sensors, temperature sensors, acoustic sensors, oil level sensors, and other optional sensors are deployed. Secondly, during the collaborative operation of the industrial robot with external loads or equipment, electronic devices can acquire different operational detection signals from the various types of multi-source sensors deployed on the industrial robot, such as speed detection signals, current detection signals, and vibration detection signals. Furthermore, after obtaining these multi-source operational detection signals, the electronic devices can perform synchronous alignment processing on all signals and use one of the operational detection signals as the reference operational detection signal. For example, if the speed detection signal is used as the reference operational detection signal, then each operational detection signal can be represented as S. i =[x i_1 x i_2 , ..., x i_D ], where i represents the sequence number of each operational detection signal, such as i=1 for speed detection signal, i=2 for current detection signal, ..., i=k for vibration detection signal; and this reference operational detection signal can be expressed as S1=[x 1_1 x 1_2 , ..., x 1_D], where D represents the total number of sampling points collected within one acquisition cycle, and optionally, it can also be the sampling point number; and x in each running detection signal is the acquired data of the running detection signal at one acquisition time. Secondly, for the value of the signal data in the reference running detection signal, if the value of the signal data is within the range of 0-300 degrees / second (° / s), that is, its value range is 0-300, then the interval range can be defined according to this value range; for example, it can be divided into 3 intervals according to the relatively even value range, namely the interval range of [0, 100), [100, 200), and [200, 300); or, for example, the interval can be reasonably divided according to the value distribution of the signal data. For example, if there are more signal data values ​​between the value of 10 and the value of 100, then the interval range of 0-100 can be refined into [0, 20), [20, 40), [40, 60), [60, 80), and [80, 100). After defining the interval ranges, we can count the number of data points within each interval based on the value of each signal data point. For example, taking the interval range [20, 40) as an example, if there are 8 signal data points within this interval range, then the frequency of the interval frequency pair corresponding to the interval range [20, 40) is determined to be 8, resulting in the interval frequency pair "[20, 40)-8". Furthermore, these 8 signal data points within this interval range, such as those corresponding to x... 1_5 -x 1_10 x 1_36 -x 1_37 Then, the position set can be determined based on the position sequence number (such as 5-10, 36-37) of these corresponding signal data in the reference operation detection signal.

[0042] In this embodiment, a baseline operational detection signal is first determined from multiple sources and different types of detection signals. This allows for preliminary signal analysis using one of the detection signals as a benchmark, providing a foundation for subsequently determining the regions of other detection signals related to the baseline operational detection signal. Simultaneously, intervals are divided based on the numerical values ​​of the signal data in the baseline operational detection signal, and the frequency of signal data within each interval is determined. This, in turn, determines the set of locations corresponding to the signal data within each interval, providing a basis for subsequently selecting signal data from related regions of other operational detection signals based on the location set determined from the baseline operational detection signal.

[0043] S102. Determine the reference signal subset corresponding to the reference operation detection signal in the interval frequency pair according to the position interval range of the position set, and determine the non-reference signal subset corresponding to each non-reference operation detection signal according to the data time corresponding to the signal data in the reference signal subset.

[0044] The location interval range represents the range of location indices within the location set. Data time refers to the acquisition time point corresponding to the signal data in the reference signal subset. Non-reference operation detection signals are operation detection signals other than the reference operation detection signals.

[0045] Specifically, after obtaining the location set, the reference signal subset corresponding to the reference operation detection signal in the interval subset can be determined based on the location interval range in the location set. For example, as mentioned above, if the location interval range in the location set is 5-10 and 36-37, then the signal data with "D" of 5-10 and 36-37 in the reference operation detection signal can be used as the reference signal subset corresponding to the interval subset. Furthermore, based on each signal data in the reference signal subset, the time point at which it was collected can be determined to obtain the data time corresponding to the signal data in the reference signal subset. Finally, the data time range is determined according to the data time corresponding to the signal data in the reference signal subset, and for each non-reference operation detection signal, the signal data corresponding to the same data time range in that non-reference operation detection signal is used as the element in the non-reference signal subset corresponding to that non-reference operation detection signal.

[0046] In this embodiment, the reference signal subset corresponding to the frequency pair of the reference operation detection signal is first determined based on the location interval range of the location set corresponding to each interval range. This enables reasonable clustering and division of the signal data in the reference operation detection signal based on the frequency pair of intervals. Then, based on the data time corresponding to the signal data in the reference signal subset, data with the same acquisition time point as the signal data in the reference signal subset is selected from each non-reference operation detection signal to form a non-reference signal subset. This allows for the selection of detection signals from other operation detection signals that have the same acquisition time point as the reference operation detection signal, using one operation detection signal as a reference. This accurately and reasonably reflects the operating status of the RV reducer during the acquisition period, providing an accurate data foundation for subsequent model building.

[0047] S103. Take the reference signal subset and all non-reference signal subsets as samples and input them into the sub-model corresponding to the interval frequency pair to obtain the model diagnosis results output by the sub-model.

[0048] In this embodiment, each interval frequency pair corresponds to a sub-model, and all sub-models are combined into a fault diagnosis model. Each sub-model has its corresponding weight coefficient.

[0049] Optionally, in this embodiment, after obtaining the above-mentioned interval frequency pairs, the frequency pairs can be filtered according to the preset number of sub-models, and each selected interval frequency pair corresponds to one sub-model. When filtering interval frequency pairs, interval frequency pairs with larger frequencies can be selected first based on the order of frequency size.

[0050] Specifically, following the steps described above, the reference signal subsets and all non-reference signal subsets corresponding to different frequency pairs can be obtained. Then, the reference signal subset and all non-reference signal subsets corresponding to a frequency pair can be used as training samples and input into the sub-model corresponding to that frequency pair in the fault diagnosis model. The sub-model trains the frequency pair samples and outputs the model's diagnostic results.

[0051] S104. Determine the overall model diagnosis result based on the model diagnosis results output by the corresponding sub-models according to the interval frequency. Determine the single accuracy coefficient based on the model diagnosis results output by each sub-model. Adjust the weight coefficient of the sub-models according to the overall model diagnosis result and the single accuracy coefficient of each sub-model to obtain the fault diagnosis model after modeling.

[0052] Specifically, after outputting the model diagnosis results for each interval frequency pair corresponding to the sub-model, the overall model diagnosis result can first be determined based on the model diagnosis results of all sub-models and their respective weights. Simultaneously, the individual accuracy coefficient is determined based on the model diagnosis results of each sub-model and the actual fault label. Finally, the weight coefficients of the sub-models can be adjusted based on the overall model diagnosis result and the individual accuracy coefficients of each sub-model. Thus, after training, the modeled fault diagnosis model can be obtained.

[0053] In this embodiment, a frequency pair within an interval is used as a benchmark. The benchmark signal subset and non-benchmark signal subset corresponding to this frequency pair are used as samples to train the sub-model corresponding to that frequency pair. By extracting signal data of the frequency pair within the interval under steady-state or quasi-steady-state operating conditions, the diagnostic model can effectively eliminate the mutual interference of external factors such as speed fluctuations, load changes, and process changes in different operating states of the RV reducer under complex industrial scenarios. This ensures that the trained sub-model has stronger adaptability to the signal data corresponding to the frequency pair after training, thereby improving the fault detection accuracy of the sub-model for that frequency pair. To obtain more reliable and accurate diagnostic results, the diagnostic results under different partitioned operating conditions are considered. The sub-models under different partitions are weighted to obtain a comprehensive fault diagnosis model, improving the robustness and diagnostic accuracy of the model and outputting more reliable results. Meanwhile, by adjusting the weight coefficients of sub-models based on the model's comprehensive diagnostic results and the single accuracy coefficient, each sub-model in the fault detection model can adaptively adjust the optimal weight coefficient ratio of the sub-model according to the detection data and the model diagnostic results of the sub-model based on the frequency of different intervals. This avoids the lack of flexibility in model integration caused by static weight allocation during the sub-model integration process, which prevents dynamic adjustment based on changes in data distribution or the characteristics of specific input samples, thus limiting the adaptability and overall performance limit of the integrated model.

[0054] The fault diagnosis modeling method provided in this invention determines a baseline operating detection signal from multiple sources and different types of detection signals, enabling preliminary signal analysis based on one of the detection signals. Simultaneously, it divides the signal data in the baseline operating detection signal into interval ranges and determines the frequency of signal data within each interval range, thereby determining the location set corresponding to the signal data in each interval range. This provides a basis for selecting signal data from related regions in other operating detection signals based on the baseline operating detection signal. Determining the baseline signal subset corresponding to the interval frequency pair based on the location interval set corresponding to each interval range allows for reasonable clustering and division of the signal data in the baseline operating detection signal using interval frequency pairs as a basis. Furthermore, based on the data time corresponding to the signal data in the baseline signal subset, data with the same acquisition time point as the signal data in the baseline signal subset is selected from each non-baseline operating detection signal to form a non-baseline signal subset. This enables the selection of detection signals from other operating detection signals that have the same acquisition time point as the baseline operating detection signal, accurately and reasonably reflecting the operating status of the RV reducer during the acquisition period, using one operating detection signal as a baseline. Using a frequency pair of intervals as a benchmark, the benchmark signal subset and non-benchmark signal subset corresponding to that frequency pair are used as samples to train the sub-model corresponding to that frequency pair. This ensures that the trained sub-model has a stronger adaptability to the signal data corresponding to that frequency pair after training, thereby improving the fault detection accuracy of the sub-model for that frequency pair. Simultaneously, by adjusting the weight coefficients of the sub-models based on the model's comprehensive diagnostic results and individual accuracy coefficients, the weight coefficient ratio of each sub-model in the fault detection model can be adjusted according to the detection data in different frequency pairs and the model's diagnostic results. This integrates the sub-models into a comprehensive fault diagnosis model, improving the model's robustness and diagnostic accuracy, and outputting more reliable results.

[0055] Figure 2 This is a flowchart illustrating another modeling method for a fault diagnosis model provided in this embodiment of the invention. This embodiment elaborates on the steps of determining interval frequency pairs, determining location sets, determining a subset of reference signals, determining a subset of non-reference signals, determining the comprehensive diagnostic result of the model, and adjusting the weight coefficients of different sub-models, based on the above embodiment. In this embodiment, the method may include:

[0056] S201. Acquire at least two operating detection signals of the rotary vector reducer collected by each multi-source sensor, and select a reference detection signal from the at least two operating detection signals based on the detection steady-state characteristics.

[0057] The multi-source sensors can include angle sensors, current sensors, vibration sensors, temperature sensors, sound sensors, oil level sensors, and other optional sensors. The rotary vector reducer is also known as an RV reducer. Detecting steady-state characteristics is used to characterize the RV reducer's stable state under different operating conditions.

[0058] Specifically, when an industrial robot starts running, multiple multi-source sensors deployed on the industrial robot can be controlled to collect their corresponding operation detection signals, resulting in at least two operation detection signals. In practice, multiple operation detection signals may be obtained, such as speed detection signals, current detection signals, and torque detection signals.

[0059] Optionally, after obtaining at least two operational detection signals, each operational detection signal can be numbered. For example, the speed detection signal can be designated as number 1, the vibration detection signal as number 2, the current detection signal as number 3, and so on. Furthermore, the signal data in each operational detection signal can be sorted according to the acquisition time sequence. Taking the speed detection signal as an example, the speed detection signal can be represented as S1=[x 1_1 x 1_2 , ..., x 1_20 That is, the speed detection signal is acquired at the first acquisition time point within one acquisition cycle. 1_1 This signal data was collected at the second acquisition time point x. 1_2 This signal data… and so on, collecting 20 signal data points at preset intervals. It's worth noting that in practice, the number of signal data points collected could be much higher; this is merely an example.

[0060] S202. Divide the signal data in the reference detection signal into at least two interval ranges, and match all the signal data in the reference detection signal with each interval range to obtain the interval frequency pair corresponding to each interval range.

[0061] The numerical range refers to the value range of all signal data values.

[0062] Specifically, in the reference detection signal, each signal data has its corresponding numerical value, such as x. 1_1 =0° / s, x 1_5 =15° / s. Therefore, based on the value range of all signal data, the numerical range can be divided into at least two intervals, each of which can be denoted as B. j Such as B1, B2, ..., B j B ZWhere j represents the j-th interval range, and Z indicates that there are a total of Z interval ranges; the lower bound of each interval range is less than its upper bound, and the lower bound of the interval range adjacent to it is greater than its upper bound. Furthermore, by matching each signal data in the reference detection signal with each interval range and counting the number of signal data falling within each interval range, the interval frequency pairs corresponding to each interval range can be obtained.

[0063] For example, suppose the signal data in the reference detection signal is S1={x 1_1 =0, x 1_2 =0, x 1_3 =7,x 1_4 =12, x 1_5 =15, x 1_6 =22, x 1_7 =26, x 1_8 =28, x 1_9 =35, x 1_10 =52, ..., x 1_20 =69}, based on the values ​​corresponding to all signal data, the value range can be determined as [0, 70) (which can be rounded to 69); furthermore, one implementation method is to divide it into at least two interval ranges more evenly, for example, into [0, 10), [10, 20), [20, 30), ..., or into [0, 35) and [35, 70). Another implementation method is to divide the range into different ranges based on the frequency of the values ​​corresponding to the signal data in a certain interval range. For example, there are many values ​​in [0, 20), so it can be subdivided based on [0, 20), such as finally into [0, 10), [10, 20), [20, 40), [40, 70).

[0064] For example, all signal data in the reference detection signal are matched with each interval range to obtain interval frequency pairs corresponding to each interval range, including:

[0065] (i) For each signal data in the reference detection signal, determine the interval range to which the signal data belongs based on the value of the signal data and each interval range.

[0066] Specifically, continuing from the example above, after dividing the data into at least two intervals, if the intervals are [0, 10), [10, 20), [20, 40), [40, 70), then for x... 1_1 =0, which determines x 1_1 This signal data belongs to the interval [0, 10); similarly, x 1_2 =0 and x 1_3 =7 also falls within the interval [0, 10), x 1_4 =12 and x1_5 =15 belongs to the interval [10, 20), x 1_6 =22, x 1_7 =26, x 1_8 =28, with x 1_9 =35 belongs to the interval [20, 40), x 1_10 =52 and x 1_20 =69 belongs to the interval [40, 70).

[0067] (ii) For each interval range, determine the number of signal data belonging to the interval range to obtain the interval frequency pair corresponding to each interval range.

[0068] Specifically, continuing from the example above, the number of signal data in the interval [0, 10) is 3, the number of signal data in the interval [10, 20) is 2, the number of signal data in the interval [20, 40) is 4, and the number of signal data in the interval [40, 70) is 2. Thus, the interval frequency pairs corresponding to each interval range can be obtained: for the interval [0, 10), the interval frequency pair is "[0, 10)-3"; for the interval [10, 20), the interval frequency pair is "[10, 20)-2"; for the interval [20, 40), the interval frequency pair is "[20, 40)-4"; and for the interval [40, 70), the interval frequency pair is "[40, 70)-2".

[0069] Optionally, in this embodiment, signal data with values ​​close to 0 (e.g., data within a pre-defined range, [0, 5)) can also be removed. One implementation involves removing data close to 0 when determining the reference detection signal, and then calculating the frequency interval subsets corresponding to different interval ranges. Another implementation involves removing signal data belonging to the interval frequency pairs within the range [0, preset value) (e.g., the region closest to 0, [0, 10)).

[0070] In this embodiment, by dividing the signal data in the benchmark operation detection signal into interval ranges and determining the interval frequency pairs, it is possible to cluster the signal data in the benchmark operation detection signal according to its numerical values, providing a data foundation for training the corresponding sub-model based on this data. Optionally, signal data with near-zero values ​​is removed, taking into account the detection signals corresponding to the stopped or quasi-static state of the industrial robot during operation, to prevent the accuracy of diagnosing fault characteristics of the RV reducer from decreasing due to signal data with near-zero values.

[0071] Optionally, after obtaining the interval frequency pairs, all interval frequency pairs corresponding to the interval range can be sorted according to the frequency in the set to obtain sorted interval frequency pairs.

[0072] Specifically, following the steps above, each interval range can be assigned a corresponding interval frequency pair. Therefore, the interval frequency pairs can be sorted according to their frequencies to obtain sorted interval frequency pairs.

[0073] For example, suppose there are three interval ranges. The frequency of the interval frequency pair corresponding to the first interval range is 8, the frequency of the interval frequency pair corresponding to the second interval range is 10, and the frequency of the interval frequency pair corresponding to the third interval range is 6. Then, in the sorted interval frequency pairs, the interval frequency pair corresponding to the second interval range is ranked first, the interval frequency pair corresponding to the first interval range is ranked second, and the interval frequency pair corresponding to the third interval range is ranked third.

[0074] Optionally, in the subsequent steps of this step in this embodiment, all the interval frequency pairs obtained above can be used to perform the following steps, or only the interval frequency pairs with the highest order can be selected to continue performing the following steps.

[0075] S203. For any interval frequency pair in the interval frequency pair, determine the first signal data belonging to the interval range based on the interval range and the value of each signal data.

[0076] Specifically, for any interval frequency pair, the first signal data belonging to that interval range can be determined based on the interval range corresponding to that interval frequency pair and the value of each signal data.

[0077] Optionally, in this embodiment, if the frequency pairs of intervals have been sorted as described above, the following steps can be performed starting from the interval frequency pair with the highest frequency, according to the sorting order.

[0078] For example, continuing from the above example, S1={x 1_1 =0, x 1_2 =0, x 1_3 =7,x 1_4 =12, x 1_5 =15, x 1_6 =22, x 1_7 =26, x 1_8 =28, x 1_9 =35, x 1_10 =52, ..., x 1_20 =69}, if the interval range is [20, 40), then the interval frequency pair is “[20, 40) - 4”, and its frequency is 4. Then the first signal data is x. 1_6 x1_7 x 1_8 x 1_9 .

[0079] S204. Determine the position set based on the data position of the first signal data in the reference operation detection signal.

[0080] Specifically, the data position of the first signal data in the reference operation detection signal can be determined directly based on the index of the first signal data, and the position set can be determined based on this.

[0081] For example, continuing from the above example, the first signal data is x 1_6 x 1_7 x 1_8 x 1_9 If the data positions are 6, 7, 8, and 9, then the set of positions is {6, 7, 8, 9}.

[0082] S205. Arrange each data position in the position set according to the position number, and determine at least one position interval range based on the consecutive number of position numbers.

[0083] Specifically, since each data location in the location set is not necessarily consecutive, it can be arranged according to its position number to facilitate finding the corresponding collection time period. For example, if the location set includes data locations 6, 7, 8, 9, 15, 18, 19, 20, 12, 13, 14, 21, 35, 36, 22, 31, 32, 23, 24, 27, 37…, it can be rearranged according to its position number to obtain the rearranged location set {6, 7, 8, 9, 12, 13, 14, 15, 18, 19, 20, 21, 22, 23, 24, 27, 31, 32, 35, 36, 37…}. Furthermore, at least one location interval can be determined based on the consecutive number of position numbers.

[0084] For example, determining at least one location range, for instance, based on the rearranged set of locations described above, allows for the determination of consecutive data locations to obtain at least one location range, such as c1=7~9, c2=12~15, c3=18~24, c4=27, c5=31~32, c6=35~37.

[0085] S206. For each position interval range, determine the reference signal subset based on the data position in each position interval range and the signal data in the reference operation detection signal.

[0086] Specifically, before determining the subset of the reference signal, the range of all position intervals can be further defined. For example, the boundary [a, b] of the position interval can be determined, and it can be determined whether a and b belong to the position set; if not, the position interval range can be redefined. All data positions within the position interval can be determined to belong to the position set; if not, the position interval range can be redefined. It can be determined whether a is the start position of the interval and b is the end position of the interval; if not, the values ​​of a and b can be redefined. Furthermore, after determining the boundary of each position interval, the position interval ranges can be filtered, such as eliminating position intervals where b-a+1 < L, retaining only those where b-a+1 ≥ L; where L is a preset quantity threshold used to limit the minimum number of elements in the position interval range. Then, by mapping each retained position interval to the signal data in the reference operation detection signal, the subset of the reference signal can be obtained: Y. rl ={x n |n∈[a,b]};that is, Y rl As a subset of the reference signal, x n This refers to the signal data in the reference operation detection signal corresponding to the data position n within the position interval range, where n belongs to the data within the range boundary from a to b. For example, if the range boundary is [18, 24], then the corresponding signal data in the reference operation detection signal is {x}. 1_18 x 1_19 x 1_20 x 1_21 x 1_22 x 1_23 x 1_24}, that is, {x 1_18 x 1_19 x 1_20 x 1_21 x 1_22 x 1_23 x 1_24 This is a subset of the reference signal.

[0087] S207. Determine the data time corresponding to the second signal data from the reference operation detection signal based on the data position of the second signal data corresponding to the location interval range of the reference signal subset.

[0088] The data time refers to the acquisition time when the second signal data was acquired. The second signal data is the signal data within the location interval range that corresponds to the location in the reference signal subset.

[0089] Specifically, because the signal data in the initial acquisition of the reference operation detection signal is preprocessed, such as by removing outliers and data close to zero, the actual acquisition time of each signal data in the final reference signal subset, arranged according to its data position, is not necessarily the same as the signal data acquired sequentially in the acquisition order. Therefore, for each second signal data corresponding to a position range in the reference signal subset, the data time corresponding to each second signal data can be determined based on its position. For example, assuming the data position of the second signal data is 13, its signal data in the reference operation detection signal can be located, and the acquisition time of that signal data during actual acquisition can be determined based on the signal data, thus obtaining the data time corresponding to the second signal data.

[0090] S208. Determine the time interval based on the data time. For each non-reference operation detection signal, take the signal data within the time interval in the non-reference operation detection signal as the data in the non-reference signal subset.

[0091] Specifically, after obtaining the data time of each second signal data corresponding to a position interval, the time interval for acquiring the second signal data within that position interval can be determined based on the data times of all the second signal data. For example, if a position interval corresponds to signal data from bit 18 to bit 24 of the reference operation detection signal, and the data time corresponding to bit 18 as second signal data is determined to be 4.50 ms after the start of acquisition, ..., and the data time corresponding to bit 24 as second signal data is 4.98 ms after the start of acquisition, then the time interval corresponding to the second signal data within that position interval can be determined to be 4.50 ms - 4.98 ms. Furthermore, signal data within this time interval can be determined from non-reference operation detection signals, and these signal data can be identified as data within the non-reference signal subset. For example, using the current detection signal as the non-reference operation detection signal, the time of acquiring signal data can be directly selected as signal data from 4.50 ms to 4.98 ms after the start of acquisition, and these signal data can be considered as data within the non-reference signal subset of the current detection signal. Based on this, the subset of non-reference signals corresponding to other non-reference operation detection signals can be obtained.

[0092] Optionally, when obtaining the reference signal subset, in order to more intuitively determine the time interval and simultaneously extract other non-reference signal subsets within the same time period based on the time interval, such as displaying time as the horizontal axis and the reference features (such as speed) of the reference signal subset as the vertical axis, and determining the time interval, then other non-reference signal subsets can be extracted simultaneously based on the time interval.

[0093] In this embodiment, the extracted steady-state or quasi-steady-state velocity data (i.e., the second signal data) and the simultaneous extraction of data from all other sensor signals (such as current, vibration, etc.) within the same time period (i.e., time interval) enable the construction of diverse training samples for subsequent multi-source sensor fault diagnosis models. This data is high-quality steady-state or quasi-steady-state data, which not only achieves high diagnostic accuracy with a simple model structure but also reduces computational overhead. Furthermore, by effectively filtering out interference from other factors under non-steady-state conditions and focusing on the data segment that best reflects the inherent health state of the RV reducer, it plays a crucial role in improving the diagnostic accuracy and reliability of the subsequent fault diagnosis model.

[0094] S209. Using the reference signal subset and all non-reference signal subsets as samples and inputting the interval frequency pairs, train and predict the corresponding sub-models to obtain the model diagnostic results output by the sub-models.

[0095] In this embodiment, a sample set corresponds to a subset of reference signals and all non-reference signal subsets within a time interval. In this embodiment, a fault diagnosis model contains multiple sub-models, each sub-model corresponding to the sample set containing the reference signal subset for processing different interval frequency pairs.

[0096] Specifically, all non-reference signal subsets corresponding to the reference signal subset within the same time interval, along with the reference signal subset itself, are considered as a sample set. The sub-model corresponding to the interval frequency of the reference signal subset is then used as the model to be trained on this sample set. The sample set is input into this sub-model, which then outputs the model diagnostic results. These results include multiple fault categories and can be represented as R. g =[R g_1 R g_2 , ..., R g_C ]; where g is the g-th sub-model and C is the number of fault categories.

[0097] S210. Based on the current weight coefficients of the sub-model corresponding to each interval frequency pair and the model diagnostic results output by the sub-model, determine the first parameter corresponding to each interval frequency pair.

[0098] In this embodiment, the initial "current weight coefficient" of each sub-model can be preset according to the historical situation of its corresponding frequency interval, or it can be evenly distributed, such as the total weight divided by the number of sub-models to obtain the sub-weight coefficient. The initial weight coefficient of each sub-model is the same.

[0099] Specifically, the current weight coefficient of the sub-model can be multiplied by its output model diagnostic result to obtain the first parameter corresponding to the interval frequency pair of the sub-model.

[0100] S211. Sum the interval frequencies with respect to the corresponding first parameter to obtain the model's comprehensive diagnostic results.

[0101] Specifically, by adding the interval frequencies to the corresponding first parameter, the comprehensive diagnostic result of the model can be obtained. That is, steps S210-S211 can be calculated according to the following formula:

[0102] ;

[0103] Where R represents the overall diagnostic result of the model, M represents the total number of sub-models, and the sub-models are numbered from 1 to M, ω g R represents the current weight coefficient corresponding to the g-th sub-model. g The model diagnostic result output by the g-th sub-model is the current weight coefficient corresponding to the g-th sub-model multiplied by the model diagnostic result ω output by that sub-model. g ×R g That is, the first parameter corresponding to the interval frequency pair of the sub-model.

[0104] S212. Determine the single accuracy coefficient based on the model diagnosis results output by each sub-model and the actual fault labels.

[0105] The actual fault label is the known actual fault category corresponding to each sample.

[0106] Specifically, for the model diagnostic results output by each sub-model, they are compared with the actual fault labels to determine the number of samples that accurately diagnose the fault category, thus obtaining the single accuracy coefficient.

[0107] S213. Determine the comprehensive accuracy coefficient based on the model's comprehensive diagnostic results and the actual fault labels, and adjust the current weight coefficient of each sub-model based on the comprehensive accuracy coefficient and the single accuracy coefficient corresponding to each sub-model.

[0108] Specifically, the diagnosed faults corresponding to all samples in the model's comprehensive diagnostic results are compared with the actual fault labels of all samples to determine the number of samples that accurately diagnose the fault category, thus obtaining the comprehensive accuracy coefficient. Furthermore, based on the comprehensive accuracy coefficient and the individual accuracy coefficients corresponding to each sub-model, new weight coefficients can be determined for each sub-model at the start of its next training round, and the current weight coefficients of that sub-model can be adjusted using these new weight coefficients.

[0109] For example, the current weight coefficients of the sub-models are adjusted based on the overall accuracy coefficient and the individual accuracy coefficients corresponding to each sub-model, including:

[0110] (i) For each sub-model, determine the target coefficient based on the comprehensive accuracy coefficient and the individual accuracy coefficient, and determine the next weight coefficient based on the individual accuracy coefficient and the target coefficient.

[0111] The target coefficient is the benchmark coefficient for the current iteration, determined by real-time competitive evaluation of the single accuracy coefficient and the comprehensive accuracy coefficient during the iterative process of adjusting the sub-model weight parameters to optimize the comprehensive fault diagnosis model, and by extreme value selection based on the advantage criterion.

[0112] Specifically, the overall accuracy coefficient can be compared with the individual accuracy coefficients, and the larger value can be selected as the target coefficient. Then, the next weight coefficient for this sub-model can be determined by dividing the individual accuracy coefficient by the target coefficient.

[0113] (ii) Use the next weighting coefficient as the current weighting coefficient.

[0114] Specifically, after calculating the next weight coefficient of the sub-model, it can be used as the current weight coefficient of the sub-model in the next round of training.

[0115] Optionally, for each sub-model, the training termination condition is when the absolute value of the difference between the next weight coefficient and the current weight coefficient is less than or equal to a preset first weight threshold; or when the training round reaches a preset training round position; or when the absolute value of the difference between the average of the next weight coefficients and the average of the current weight coefficients for all sub-models is less than or equal to a preset second weight threshold.

[0116] Optionally, for more complex multi-operating conditions, these conditions can be decomposed into multiple single-operating conditions. Then, for each single-operating condition, the multi-interval signals can be extracted using steps S201-S208 as described above. Next, the weights of each sub-model are optimized and allocated using steps S209-S213 as described above, finally yielding the comprehensive fault diagnosis model for the RV reducer. When more new samples are acquired throughout the RV reducer's lifespan, the model can be updated in real-time for fault diagnosis and maintenance. When suitable new samples are added to the sample library, steady-state or quasi-steady-state data can be extracted using steps S201-S208. Then, the weights of the current fault diagnosis sub-model are updated using steps S209-S213 as described above to complete the update of the new comprehensive fault diagnosis model. Based on the updated comprehensive fault diagnosis model, the RV reducer is diagnosed online in real-time for faults. If the fault is severe, the relevant components need to be replaced immediately; if the fault is not severe, predictive maintenance can be performed, inventory management can be implemented in advance, and then scheduled maintenance of the RV reducer can be carried out as planned.

[0117] Optionally, during the actual operation of an industrial robot, the RV reducer may operate at different speeds after being driven by the motor. Under complex working conditions, the relatively steady-state or quasi-steady-state (speed variation within a certain set small range) operating speed data exhibits periodicity and stability. Issues such as imbalance, misalignment, broken gear teeth, and bearing wear in the RV reducer will generate specific, repetitive, and clear characteristic frequencies in the steady-state signal. In steady-state or quasi-steady-state data, fault signal characteristics benefit from a higher signal-to-noise ratio, allowing for better separation from the strong background noise of the industrial environment. Furthermore, compared to complex speed-changing situations, the fault characteristic frequencies of steady-state or quasi-steady-state signals do not drift drastically due to speed changes. Therefore, choosing speed characteristics as the benchmark operating detection signal not only provides a simpler and more reliable basis for the subsequent fault diagnosis model but also effectively avoids significant degradation in the accuracy of RV reducer fault diagnosis in complex industrial operating scenarios. Simultaneously, in steady-state or quasi-steady-state conditions, interference from external factors such as load changes and speed fluctuations on the fault signal can be effectively eliminated, ensuring high reliability and consistency of the multi-source sensor data used in each diagnosis. In RV reducer fault diagnosis, the reliability and consistency of the diagnostic results are very important, as this ensures the accuracy and predictability of diagnosis and maintenance.

[0118] Regular monitoring of steady-state or quasi-steady-state characteristics or diagnostic results allows for continuous tracking of the amplitude changes of fault characteristics or diagnostic results over time. Based on these trends, the remaining usability of the RV reducer can be accurately predicted. This enables maintenance personnel to promptly identify and resolve potential problems before minor faults escalate into serious accidents, and allows for advance planning of inventory management and scheduled maintenance, avoiding untimely downtime caused by delayed maintenance.

[0119] Optionally, in this embodiment, the industrial robot will work collaboratively with external loads or equipment. During this collaborative operation, various types of multi-source sensors deployed externally or internally to the industrial robot can collect data on the operation and status of the RV reducer. Each sensor can also feed back the periodically collected data to the electronic device in this embodiment. Furthermore, to facilitate data understanding and analysis, the electronic device can align the timestamps of the sensor data collected by each sensor to obtain periodically collected data. Simultaneously, it can perform necessary signal amplification, filtering and noise reduction, data cleaning, and data standardization preprocessing on the collected signals. Based on the processed sensor data and the pre-trained sub-models, the current health status and remaining available time of the RV reducer can be determined. Furthermore, the electronic device can be configured with a fault alarm unit and a maintenance planning unit. The fault alarm unit determines the health status of the RV reducer based on the fault diagnosis results and issues a fault alarm. The maintenance planning unit predicts maintenance based on the remaining available time and provides maintenance suggestions such as the optimal maintenance time. This allows maintenance personnel to rationally allocate the cycles of purchasing, repairing, and replacing RV reducers, maximizing RV reducer utilization while minimizing the risk of industrial robot production downtime, extending component lifespan, and reducing maintenance costs. Additionally, the electronic device can also be configured with an operation strategy optimization unit, which appropriately adjusts the industrial robot's work processes and operating conditions based on the current diagnostic results of the RV reducer and the production strategy.

[0120] Figure 3 This is a schematic diagram of a modeling device for a fault diagnosis model provided in an embodiment of the present invention. Figure 3 As shown, the device includes:

[0121] The first determining module 301 is used to determine interval frequency pairs based on the value and interval range of signal data in the reference operation detection signal, and for any interval frequency pair, determine the position set based on the interval range and signal data corresponding to the interval frequency pair.

[0122] The second determining module 302 is used to determine the reference signal subset corresponding to the reference operation detection signal in the interval frequency pair according to the position interval range of the position set, and to determine the non-reference signal subset corresponding to each non-reference operation detection signal according to the data time corresponding to the signal data in the reference signal subset.

[0123] The output module 303 is used to take the reference signal subset and all non-reference signal subsets as samples and input the interval frequency pairs to train and predict the corresponding sub-models, so as to obtain the model diagnosis results output by the sub-models.

[0124] The adjustment module 304 is used to determine the overall model diagnosis result based on the model diagnosis result output by the corresponding sub-model according to the interval frequency, determine the single accuracy coefficient based on the model diagnosis result output by each sub-model, and adjust the weight coefficient of the sub-model according to the overall model diagnosis result and the single accuracy coefficient of each sub-model.

[0125] Based on the above embodiments, the first determining module 301 is specifically used to determine interval frequency pairs according to the numerical value and interval range of the signal data in the reference operation detection signal.

[0126] Acquire at least two operating detection signals of the rotary vector reducer collected by various multi-source sensors, and select a reference detection signal from the at least two operating detection signals according to the detection steady-state characteristics; divide at least two interval ranges according to the numerical range of the signal data in the reference detection signal, and match all signal data in the reference detection signal with each interval range to obtain the interval frequency pair corresponding to each interval range.

[0127] Based on the above embodiments, all signal data in the reference detection signal are matched with each interval range to obtain interval frequency pairs corresponding to each interval range. The first determining module 301 is specifically used for:

[0128] For each signal data in the reference detection signal, the interval range to which the signal data belongs is determined based on the value of the signal data and each interval range; for each interval range, the number of signal data belonging to the interval range is determined, and the interval frequency pair corresponding to each interval range is obtained.

[0129] Based on the above embodiments, the first determining module 301 is specifically used to determine the location set according to the interval frequency and the corresponding interval range and signal data.

[0130] The first signal data belonging to the interval range is determined based on the interval range and the value of each signal data; the position set is determined based on the data position of the first signal data in the reference operation detection signal.

[0131] Based on the above embodiments, the second determining module 302 is specifically used to determine the subset of reference signals corresponding to the interval frequency pairs of the reference operation detection signal according to the position interval range of the position set.

[0132] Arrange each data position in the position set according to its position number, and determine at least one position interval range based on the consecutive number of position numbers; for each position interval range, determine a reference signal subset based on each data position in the position interval range and the signal data in the reference operation detection signal.

[0133] Based on the above embodiments, the second determining module 302 is specifically used to determine the non-reference signal subset corresponding to each non-reference operation detection signal according to the data time corresponding to the signal data in the reference signal subset.

[0134] Based on the data position of the second signal data corresponding to the location interval range in the reference signal subset, the data time corresponding to the second signal data is determined from the reference operation detection signal; based on the data time, the time interval is determined, and for each non-reference operation detection signal, the signal data in the non-reference operation detection signal that falls within the time interval is taken as the data in the non-reference signal subset.

[0135] Based on the above embodiments, the comprehensive model diagnosis result is determined according to the model diagnosis results output by the corresponding sub-models based on the interval frequency. The adjustment module 304 is specifically used for:

[0136] Based on the current weight coefficients of the sub-model corresponding to each interval frequency pair and the model diagnostic results output by the sub-model, determine the first parameter corresponding to each interval frequency pair; sum the first parameters corresponding to the interval frequency pairs to obtain the comprehensive model diagnostic results.

[0137] Based on the above embodiments, the single accuracy coefficient is determined according to the model diagnostic results output by each sub-model, and the weight coefficient of each sub-model is adjusted according to the overall model diagnostic results and the single accuracy coefficient of each sub-model. The adjustment module 304 is specifically used for:

[0138] The single accuracy coefficient is determined based on the model diagnosis results and actual fault labels output by each sub-model; the comprehensive accuracy coefficient is determined based on the comprehensive model diagnosis results and actual fault labels; and the current weight coefficients of the sub-models are adjusted based on the comprehensive accuracy coefficient and the single accuracy coefficients corresponding to each sub-model.

[0139] Based on the above embodiments, the current weight coefficients of the sub-models are adjusted according to the overall accuracy coefficient and the individual accuracy coefficients corresponding to each sub-model. The adjustment module 304 is specifically used for:

[0140] For each sub-model, the target coefficient is determined based on the overall accuracy coefficient and the individual accuracy coefficient, and the next weight coefficient is determined based on the individual accuracy coefficient and the target coefficient; the next weight coefficient is used as the current weight coefficient.

[0141] The fault diagnosis modeling device provided in the embodiments of the present invention can execute the fault diagnosis modeling method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0142] It is worth noting that in the embodiments of the above-mentioned fault diagnosis model modeling device, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the scope of protection of the present invention.

[0143] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Figure 4 A block diagram is shown of an exemplary electronic device 11 suitable for implementing embodiments of the present invention. Figure 4 The electronic device 11 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0144] like Figure 4 As shown, the electronic device 11 is represented in the form of a general-purpose computing electronic device. The components of the electronic device 11 may include, but are not limited to: one or more processors or processing units 16, system memory 28, and bus 18 connecting different system components (including system memory 28 and processing unit 16).

[0145] Bus 18 represents one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. For example, these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0146] Electronic device 11 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by electronic device 11, including volatile and non-volatile media, removable and non-removable media.

[0147] System memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. Electronic device 11 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 34 may be used to read and write non-removable, non-volatile magnetic media (… Figure 4 Not shown; usually referred to as a "hard drive"). Although Figure 4As not shown, disk drives for reading and writing to removable non-volatile disks (e.g., "floppy disks") and optical disc drives for reading and writing to removable non-volatile optical discs (e.g., CD-ROMs, DVD-ROMs, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 via one or more data media interfaces. System memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of the present invention.

[0148] A program / utility 40 having a set (at least one) of program modules 46 may be stored, for example, in system memory 28. Such program modules 46 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 46 typically perform the functions and / or methods described in the embodiments of the present invention.

[0149] Electronic device 11 can also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), and with one or more devices that enable a user to interact with electronic device 11, and / or with any device that enables electronic device 11 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed via input / output (I / O) interface 22. Furthermore, electronic device 11 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 20. Figure 4 As shown, network adapter 20 communicates with other modules of electronic device 11 via bus 18. It should be understood that, although... Figure 4 As not shown, other hardware and / or software modules may be used in conjunction with electronic device 11, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0150] The processing unit 16 executes various functional applications and page displays by running programs stored in the system memory 28, such as implementing the fault diagnosis modeling method provided in this embodiment. Of course, those skilled in the art will understand that the processor can also implement the technical solutions of the fault diagnosis modeling method provided in any embodiment of the present invention.

[0151] This invention provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, the program implements a modeling method for a fault diagnosis model, as provided in this invention. The computer storage medium of this invention can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0152] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0153] Program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0154] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements a modeling method for a fault diagnosis model as provided in any embodiment of this invention.

[0155] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0156] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computing device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0157] Furthermore, the acquisition, storage, use, and processing of data in the technical solution of this invention all comply with the relevant provisions of national laws and regulations.

[0158] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.

Claims

1. A modeling method for a fault diagnosis model, characterized in that, The method includes: Based on the numerical value and interval range of the signal data in the baseline operation detection signal, interval frequency pairs are determined. For each interval frequency pair, a set of positions is determined based on the interval range corresponding to the interval frequency pair and the signal data. Based on the location interval range of the location set, determine the reference signal subset corresponding to the reference operation detection signal in the interval frequency pair, and determine the non-reference signal subset corresponding to each non-reference operation detection signal based on the data time corresponding to the signal data in the reference signal subset; The reference signal subset and all non-reference signal subsets are used as samples and input into the sub-model corresponding to the interval frequency pair to obtain the model diagnosis result output by the sub-model; The overall model diagnosis result is determined based on the model diagnosis results output by the corresponding sub-models according to the frequency of the interval. The single accuracy coefficient is determined based on the model diagnosis results output by each sub-model. The weight coefficients of the sub-models are adjusted according to the overall model diagnosis result and the single accuracy coefficients of each sub-model to obtain the modeled fault diagnosis model.

2. The method according to claim 1, characterized in that, The step of determining the interval frequency pair based on the numerical value and interval range of the signal data in the reference operation detection signal includes: Acquire at least two operating detection signals of the rotary vector reducer collected by various multi-source sensors, and select the reference detection signal from the at least two operating detection signals based on the detection steady-state characteristics; Divide the signal data in the reference detection signal into at least two interval ranges, and match all the signal data in the reference detection signal with each interval range to obtain interval frequency pairs corresponding to each interval range.

3. The method according to claim 2, characterized in that, The step of matching all signal data in the reference detection signal with each interval range to obtain interval frequency pairs corresponding to each interval range includes: For each signal data in the reference detection signal, the interval range to which the signal data belongs is determined based on the value of the signal data and each interval range; For each interval range, determine the number of signal data belonging to the interval range to obtain the interval frequency pair corresponding to each interval range.

4. The method according to claim 1, characterized in that, The step of determining the location set based on the interval frequency, the corresponding interval range, and the signal data includes: The first signal data belonging to the interval range is determined based on the interval range and the value of each signal data; The location set is determined based on the data position of the first signal data in the reference operation detection signal.

5. The method according to claim 1, characterized in that, The step of determining the subset of reference signals corresponding to the reference operation detection signal in the frequency pair of the interval based on the location interval range of the location set includes: Arrange each data location in the location set according to its location number, and determine at least one location interval range based on the consecutive number of the location numbers. For each of the aforementioned position intervals, a subset of the reference signal is determined based on each data position within the aforementioned position interval and the signal data in the reference running detection signal.

6. The method according to claim 5, characterized in that, The step of determining the non-reference signal subset corresponding to each non-reference operation detection signal based on the data time corresponding to the signal data in the reference signal subset includes: The data time corresponding to the second signal data is determined from the reference operation detection signal based on the data position of the second signal data corresponding to the location interval range of the reference signal subset. Based on the data time, a time interval is determined. For each non-reference operation detection signal, the signal data within the time interval of the non-reference operation detection signal is taken as data in the non-reference signal subset.

7. The method according to claim 1, characterized in that, The step of determining the comprehensive model diagnosis result based on the model diagnosis results output by the corresponding sub-models according to the interval frequency includes: Based on the current weight coefficient of the sub-model corresponding to each frequency interval pair and the model diagnostic result output by the sub-model, determine the first parameter corresponding to each frequency interval pair; The comprehensive diagnostic result of the model is obtained by summing the interval frequencies with respect to the corresponding first parameter.

8. The method according to claim 7, characterized in that, The step of determining the single accuracy coefficient based on the model diagnostic results output by each sub-model, and adjusting the weight coefficients of the sub-models based on the overall model diagnostic results and the single accuracy coefficients of each sub-model, includes: The single accuracy coefficient is determined based on the model diagnostic results output by each sub-model and the actual fault label; The overall accuracy coefficient is determined based on the comprehensive diagnostic results of the model and the actual fault labels, and the current weight coefficient of each sub-model is adjusted based on the overall accuracy coefficient and the single accuracy coefficient corresponding to each sub-model.

9. The method according to claim 8, characterized in that, The step of adjusting the current weight coefficients of the sub-models based on the overall accuracy coefficient and the individual accuracy coefficients corresponding to each sub-model includes: For each of the sub-models, a target coefficient is determined based on the overall accuracy coefficient and the individual accuracy coefficient, and the next weight coefficient is determined based on the individual accuracy coefficient and the target coefficient. Use the next weight coefficient as the current weight coefficient.

10. A modeling apparatus for a fault diagnosis model, characterized in that, The device includes: The first determining module is used to determine interval frequency pairs based on the value and interval range of signal data in the reference operation detection signal, and to determine a set of positions based on the interval range corresponding to the interval frequency pairs and the signal data for the interval frequency pairs. The second determining module is used to determine the reference signal subset corresponding to the reference operation detection signal in the interval frequency pair according to the position interval range of the position set, and to determine the non-reference signal subset corresponding to each non-reference operation detection signal according to the data time corresponding to the signal data in the reference signal subset; The output module is used to take the reference signal subset and all non-reference signal subsets as samples and input them into the sub-model corresponding to the interval frequency pair to obtain the model diagnosis result output by the sub-model; The adjustment module is used to determine the comprehensive model diagnosis result based on the model diagnosis results output by the corresponding sub-models according to the frequency of the interval, determine the single accuracy coefficient based on the model diagnosis results output by each sub-model, and adjust the weight coefficient of the sub-model according to the comprehensive model diagnosis result and the single accuracy coefficient of each sub-model to obtain the modeled fault diagnosis model.

11. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the modeling method of the fault diagnosis model as described in any one of claims 1 to 9.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the modeling method for the fault diagnosis model as described in any one of claims 1 to 9.

Images