Fault location methods, devices, electronic equipment and operating machinery

Abstract

This invention relates to the field of fault detection, and provides a fault location method, device, electronic equipment, and operating machinery. The method includes: acquiring equipment operating data of a target object; the equipment operating data includes a multivariate time series composed of various equipment operating parameters; extracting features from the multivariate time series to obtain a feature vector set; and performing hierarchical fault location on the feature vector set to obtain a fault location result. The fault location method provided by this invention can simultaneously analyze and process multiple equipment operating parameters, and by extracting features from the multivariate time series, it can obtain the correlation between operating parameters of different types of equipment, resulting in higher accuracy and reliability of the fault location result.

Description

Technical Field

[0001] This invention relates to the field of fault detection technology, and in particular to a fault location method, device, electronic equipment, and operating machinery. Background Technology

[0002] Currently, the operational safety of machinery composed of multiple systems and components, such as large industrial equipment and operational machinery, is receiving increasing attention. Accurate and timely fault location is a key factor in ensuring the safe operation of such machinery. Traditional fault location methods using machine learning are gaining popularity compared to manual fault detection methods due to their more intelligent and convenient implementation process.

[0003] However, existing machine learning-based fault location methods can only analyze and process the operating parameters of a certain type of equipment. Because they cannot obtain the correlation between the operating parameters of different types of equipment, the accuracy and reliability of the fault location results are low. Summary of the Invention

[0004] This invention provides a fault location method, device, electronic equipment, and operating machinery to address the shortcomings of existing machine learning-based fault location methods, which can only analyze and process the operating parameters of a single type of equipment and cannot obtain the correlation between operating parameters of different types of equipment, resulting in low accuracy and reliability of the fault location results. This invention achieves accurate and efficient fault location.

[0005] In a first aspect, the present invention provides a fault location method, the method comprising:

[0006] Acquire the device operation data of the target object; wherein, the device operation data includes a multivariate time series composed of various device operation parameters;

[0007] Feature extraction is performed on the multivariate time series to obtain a feature vector set;

[0008] The feature vector set is used to perform hierarchical fault location to obtain the fault location result.

[0009] According to the fault location method provided by the present invention, the step of extracting features from the multivariate time series to obtain a feature vector set includes:

[0010] The time series in the multivariate time series are arranged and combined to obtain multiple sequence subsets;

[0011] Feature extraction is performed on each of the aforementioned sequence subsets to obtain multiple feature vectors;

[0012] The feature vector set is constructed based on multiple feature vectors.

[0013] According to the fault location method provided by the present invention, the step of extracting features from each of the sequence subsets to obtain multiple feature vectors includes:

[0014] Each of the aforementioned sequence subsets is transformed by a random convolution kernel to generate multiple feature maps;

[0015] A predetermined number of feature values ​​are extracted from each of the feature maps to obtain the feature values ​​corresponding to each of the sequence subsets.

[0016] Based on the feature values ​​corresponding to each of the sequence subsets, feature vectors corresponding to each of the sequence subsets are constructed respectively, resulting in multiple feature vectors.

[0017] According to the fault location method provided by the present invention, the step of performing hierarchical fault location on the feature vector set to obtain fault location results includes:

[0018] The feature vector set is input into a pre-constructed hierarchical localization model to obtain the fault localization result;

[0019] The hierarchical localization model is obtained by training a multi-level classifier based on feature vector set samples and the corresponding fault labels at each level of the feature vector set samples.

[0020] According to the fault location method provided by the present invention, the hierarchical location model includes a first-level classifier, a second-level classifier, and a third-level classifier;

[0021] The step of inputting the feature vector set into a pre-constructed hierarchical localization model to obtain fault localization results includes:

[0022] The feature vector set is input into a first-level classifier to obtain the first-level localization result corresponding to the feature vector set;

[0023] If the first-level localization result of the feature vector set is a fault, then the feature vector set is input into the second-level classifier to obtain the second-level localization result corresponding to the feature vector set;

[0024] Based on the secondary localization result, the feature vector set is input into the corresponding tertiary classifier to obtain the tertiary localization result corresponding to the feature vector set;

[0025] The first-level positioning result, the second-level positioning result, and the third-level positioning result are used as the fault location result.

[0026] According to the fault location method provided by the present invention, the primary classifier includes multiple individual classifiers;

[0027] The step of inputting the feature vector set into a first-level classifier to obtain the first-level localization result corresponding to the feature vector set includes:

[0028] Each feature vector in the feature vector set is input into the corresponding individual classifier to obtain the prediction results of each individual classifier.

[0029] The primary localization result is determined based on the prediction results of each individual classifier.

[0030] According to the fault location method provided by the present invention, the step of acquiring the equipment operation data of the target object includes:

[0031] Obtain the raw data of the target object;

[0032] If it is determined that there are missing values ​​in the original data, the missing values ​​are filled in to obtain the device operation data.

[0033] Secondly, the present invention also provides a fault location device, the device comprising:

[0034] The acquisition module is used to acquire the device operation data of the target object; wherein, the device operation data includes a multivariate time series composed of various device operation parameters;

[0035] The first processing module is used to extract features from the multivariate time series to obtain a feature vector set;

[0036] The second processing module is used to perform hierarchical fault location on the feature vector set to obtain fault location results.

[0037] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the aforementioned fault location methods.

[0038] Fourthly, the present invention also provides a working machine that uses any of the above-described fault location methods or includes the above-described fault location device.

[0039] The fault location method, device, electronic equipment, and operating machinery provided by this invention acquire equipment operation data, including a multivariate time series composed of various equipment operating parameters, extract features from the multivariate time series to obtain a feature vector set, and finally perform hierarchical fault location on the feature vector set to obtain the fault location result. Since the multivariate time series involves multiple equipment operating parameters, multiple equipment operating parameters can be analyzed and processed simultaneously. Furthermore, feature extraction from the multivariate time series can obtain the correlation between operating parameters of different types of equipment, making the fault location result more accurate and reliable. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0041] Figure 1 This is a flowchart illustrating the fault location method provided by the present invention;

[0042] Figure 2 This is a schematic diagram of a multivariate time series in an embodiment of the present invention;

[0043] Figure 3 This is a schematic diagram of multivariate time series and sequence subsets in an embodiment of the present invention;

[0044] Figure 4 This is a schematic diagram of the system-level positioning tag and component-level positioning tag corresponding to the target object being the operating machinery in this embodiment of the invention;

[0045] Figure 5 This is a schematic diagram of the normal sample set and the faulty sample set in an embodiment of the present invention;

[0046] Figure 6 This is a schematic diagram of the tree-like fault location process in an embodiment of the present invention;

[0047] Figure 7 This is a schematic diagram of the structure of the first-level classifier in an embodiment of the present invention;

[0048] Figure 8 This is a schematic diagram of the training set data architecture used when training classifiers at each level in an embodiment of the present invention;

[0049] Figure 9 This is a schematic diagram of the data results of the cross-validation experiment in an embodiment of the present invention;

[0050] Figure 10 This is a schematic diagram of the fault location device provided by the present invention;

[0051] Figure 11 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0053] The following is combined with Figures 1 to 11 This invention describes the fault location method, apparatus, electronic equipment, and operating machinery provided in the embodiments of the present invention.

[0054] Figure 1 This invention illustrates a fault location method provided by an embodiment of the present invention, the method comprising:

[0055] Step 101: Obtain the device operation data of the target object; wherein, the device operation data includes a multivariate time series composed of various device operation parameters;

[0056] Step 102: Extract features from the multivariate time series to obtain a feature vector set;

[0057] Step 103: Perform hierarchical fault location on the feature vector set to obtain the fault location result.

[0058] In this embodiment, the target object refers to machinery composed of multiple systems and components. It can be operational machinery or industrial equipment. In this embodiment, the target object is operational machinery. The execution subject of the fault location method provided in this embodiment can be the vehicle controller on the operational machinery. Specifically, it can be implemented through a pre-trained fault location model in the vehicle controller. The fault location model can be divided into two parts: a feature extraction model and a hierarchical location model.

[0059] Equipment operation data can be obtained from the data uploaded by various devices on the working machinery through the vehicle controller. In this embodiment, the devices on the working machinery can be working equipment such as engines, hydraulic main pumps and hydraulic main valves, or data acquisition devices such as temperature sensors in the fuel tank. The equipment operation data can contain a multivariate time series composed of various equipment operation parameters.

[0060] Figure 2 A multivariate time series is presented in the form of a multivariate time series data plot. This multivariate time series involves four equipment operating parameters: equipment operating parameter 1, equipment operating parameter 2, equipment operating parameter 3, and equipment operating parameter 4. This multivariate time series can characterize the changes of various equipment operating parameters over time. Figure 2The horizontal axis represents time, and the vertical axis represents the parameter values ​​for each type of equipment operation. In practical applications, the vertical axis can also represent the health indicator values ​​corresponding to various equipment operation parameters; the specific values ​​can be set according to actual needs.

[0061] In an exemplary embodiment, acquiring device operation data of the target object may specifically include:

[0062] Obtain the raw data of the target object;

[0063] If the original data is found to contain missing values, these missing values ​​are filled in to obtain the device operation data.

[0064] Before the feature extraction step, this embodiment can first fill in missing values ​​in the original data of the target object to improve the integrity of the data. Specifically, interpolation can be used to fill in the data to ensure that each multivariate time series has no missing values.

[0065] In an exemplary embodiment, feature extraction is performed on a multivariate time series to obtain a feature vector set, which may specifically include:

[0066] By arranging and combining the time series in a multivariate time series, multiple subsets of the series can be obtained;

[0067] Feature extraction is performed on each subset of sequences to obtain multiple feature vectors;

[0068] A feature vector set is constructed based on multiple feature vectors.

[0069] In this embodiment, the multivariate time series is assumed to be an n-dimensional time series, meaning the dimension of the multivariate time series is n. After permutation and combination, W sequence subsets can be obtained. The relationship between the dimension n of the multivariate time series and the number W of sequence subsets is as follows:

[0070]

[0071] Where W is the number of sequence subsets, n is the dimension of the multivariate time series, and i is a positive integer greater than or equal to 1 and less than or equal to n.

[0072] See appendix Figure 3 Taking a multivariate time series with a dimension n of 4 as an example, meaning the multivariate time series includes four categories of equipment operating parameters: equipment operating parameter 1, equipment operating parameter 2, equipment operating parameter 3, and equipment operating parameter 4. The specific calculations for permuting and combining these multivariate time series are as follows:

[0073]

[0074] correspond Figure 3As shown, the 15 sequence subsets obtained include four unary sequence subsets, six binary sequence subsets, four ternary sequence subsets, and one quaternary sequence subset. The unary sequence subsets refer to... Figure 3 In the numbers 1, 2, 3, and 4, the binary sequence subsets refer to... Figure 3 The combinations of 1 and 2, 1 and 3, 1 and 4, 2 and 3, 2 and 4, and 3 and 4 are included. The subset of the ternary sequence refers to the combinations of 1, 2 and 3, 1, 2 and 4, 1, 3 and 4, and 2, 3 and 4. The subset of the quaternary sequence refers to the combinations of 1, 2, 3 and 4.

[0075] Since the sequence subset is obtained by arranging and combining the time series in the multivariate time series, the extracted features can reflect the correlation between various equipment operating parameters after the sequence subset is subsequently extracted, which can improve the accuracy of fault location results.

[0076] Furthermore, feature extraction is performed on each subset of sequences to obtain multiple feature vectors, which may include:

[0077] Each sequence subset is transformed by a random convolution kernel to generate multiple feature maps;

[0078] A predetermined number of feature values ​​are extracted from each feature map to obtain the feature values ​​corresponding to each sequence subset.

[0079] Based on the feature values ​​corresponding to each sequence subset, feature vectors corresponding to each sequence subset are constructed to obtain multiple feature vectors.

[0080] Still with Figure 3 Taking the four-element time series as an example, since 15 sequence subsets can be obtained by arranging and combining each time series, 15 feature vectors can be obtained by extracting features from each sequence subset.

[0081] In this embodiment, the feature extraction method employs random convolution kernel transformation. Random convolution kernel transformation can convert both univariate sequence subsets and multivariate sequence subsets into feature vectors. Each convolution kernel performs a convolution operation on a sequence subset to generate a corresponding feature map, for an n-ary time series X = (x1, x2, ..., x...). n For example, in a multivariate time series, the lengths of each time series are equal, that is, the multivariate time series are time series of equal length. In this embodiment, the length of each time series in the multivariate time series is set to m.

[0082] Among the W sequence subsets obtained after permutation and combination, taking a certain sequence subset as an example, the data x at the i-th position... i The formula for the dot product operation with the convolution kernel ω is:

[0083]

[0084] Where * denotes the dot product operation, d represents the expansion coefficient, and l kernel This represents the length of the convolution kernel ω, i.e., the number of layers in the convolution kernel ω. kernel 1 point, ω j This represents the j-th point of the convolution kernel ω.

[0085] Each convolutional kernel generates a feature map corresponding to each sequence subset by performing a sliding dot product on each sequence subset. If the number of convolutional kernels is K, then K feature maps can be obtained. By extracting a preset number of feature values ​​from each feature map, the feature values ​​corresponding to the sequence subset can be obtained.

[0086] In this embodiment, two feature values ​​are extracted from each feature map: the maximum value in the feature map and the proportion of positive values ​​in the feature map. In this way, each sequence subset can obtain 2K feature values ​​through feature extraction. The 2K feature values ​​are used to construct a feature vector, thereby converting each sequence subset into a corresponding feature vector.

[0087] Each feature vector represents a view. In this way, after feature extraction from a multivariate time series, W views can be obtained, and finally a feature vector set containing W views can be constructed.

[0088] In an exemplary embodiment, hierarchical fault location is performed on the feature vector set to obtain fault location results, which may specifically include:

[0089] The feature vector set is input into a pre-built hierarchical localization model to obtain the fault localization result;

[0090] Among them, the hierarchical localization model is obtained by training a multi-level classifier based on feature vector set samples and the corresponding fault labels at each level of the feature vector set samples.

[0091] In this embodiment, the hierarchical localization model includes multiple classifiers, which can be set according to the number of fault localization levels. In this embodiment, three fault localization levels are set: fault detection, system-level localization, and component-level localization. Accordingly, the hierarchical localization model includes a first-level classifier, a second-level classifier, and a third-level classifier. The entire hierarchical localization model can be trained using feature vector set samples and the labels of the three levels corresponding to the feature vector set samples.

[0092] In this embodiment, the feature vector set sample can be obtained by extracting features from the device operation data sample. The feature vector set sample consists of a normal sample set and a fault sample set. Each sample data in the feature vector set sample has three levels of labels, namely fault detection label, system-level positioning label and component-level positioning label.

[0093] Fault detection labels include "normal" and "faulty," used to characterize whether a fault has occurred in the sample data; system-level location labels are used to characterize the system that has experienced a fault, see appendix. Figure 4 Taking the target object as a piece of machinery as an example, a system-level positioning tag can contain S systems, such as a power system, hydraulic system, mechanical system, and electrical system; a component-level positioning tag is used to characterize the component that has failed. Figure 4 T components are shown, such as the engine, fuel tank, and radiator in the power system; the hydraulic main pump, hydraulic main valve, and boom cylinder in the hydraulic system; the bucket, stick, and traveling device in the mechanical system; and the battery, generator, and speed sensor in the electrical system. Each sample data has only one label at each level.

[0094] See appendix Figure 5 The normal sample set includes X normal samples, each with a fault detection label, system-level location label, and component-level location label all labeled "normal". The fault sample set includes Y fault samples, each with a fault detection label of "fault", and system-level and component-level location labels indicating the location of the fault, i.e., the specific faulty system or component.

[0095] To achieve fault location based on equipment operation data, a mapping function from feature vector set to predicted label is needed. In this embodiment, the classifier is trained on feature vector set samples generated from equipment operation data samples, that is, the function of the above mapping function is achieved through the trained hierarchical localization model.

[0096] Furthermore, the hierarchical localization model can include a first-level classifier, a second-level classifier, and a third-level classifier;

[0097] By inputting the feature vector set into a pre-built hierarchical localization model, the fault localization results are obtained, including:

[0098] The feature vector set is input into the first-level classifier to obtain the first-level localization result corresponding to the feature vector set;

[0099] If the first-level localization result of the feature vector set is a fault, then the feature vector set is input into the second-level classifier to obtain the second-level localization result corresponding to the feature vector set;

[0100] Based on the secondary localization results, the feature vector set is input into the corresponding tertiary classifier to obtain the tertiary localization results corresponding to the feature vector set;

[0101] The first-level, second-level, and third-level positioning results are used as the fault location results.

[0102] In this embodiment, the fault location process achieved through the hierarchical location model can be achieved by... Figure 6The tree-like fault location process shown is implemented by first inputting the feature vector set into the first-level classifier, and then outputting the first-level location result after fault detection, i.e., whether it is normal or faulty. The feature vector set that is detected as faulty will be input into the second-level classifier. After the second-level classifier performs system-level location, it outputs the second-level location result, i.e. the faulty system. In this embodiment, the faulty system can be one of system 1 to system S.

[0103] After identifying the faulty system, the corresponding feature vector can be input into the three-level classifier. After the three-level classifier performs component-level localization, it outputs the three-level localization result, i.e., the faulty component. In this embodiment, the faulty component can be determined from multiple corresponding components under the faulty system. In this embodiment, the faulty component output by the three-level classifier can be component 1 to component j.

[0104] pass Figure 6 As can be seen, this embodiment uses a tree-like fault location process to achieve fault location, which can realize the sequential execution of fault detection, system-level location and component-level location, resulting in higher fault location accuracy.

[0105] Furthermore, a first-level classifier can include multiple individual classifiers;

[0106] The feature vector set is input into the first-level classifier to obtain the first-level localization result corresponding to the feature vector set, including:

[0107] Each feature vector in the feature vector set is input into the corresponding individual classifier to obtain the prediction results of each individual classifier.

[0108] Based on the prediction results of each individual classifier, the primary localization result is determined.

[0109] In this embodiment, after obtaining the prediction results of each individual classifier, the final first-level localization result can be obtained through weighted voting.

[0110] In this embodiment, each level of classifier can include multiple individual classifiers. Taking the first-level classifier as an example, see Appendix. Figure 7 The first-level classifier consists of W individual classifiers. The W feature vectors in the feature vector set are input into the corresponding individual classifiers to obtain the fault detection results corresponding to each feature vector. Finally, all the fault detection results are weighted and voted to obtain the final output result of the classifier.

[0111] The training set data samples used when training the classifiers at each level can be found in [reference needed]. Figure 8The first-level classifier is used for fault detection. It takes a set of feature vector samples as input and outputs a first-level localization result, i.e., a fault detection label, such as "normal" or "fault". The first-level classifier is trained on a normal sample set and a fault sample set. The samples in the normal sample set are all labeled "normal", and the samples in the fault sample set are all labeled "fault".

[0112] The secondary classifier is a system-level localization classifier. It takes the detection results from the input feature vector set as the fault feature vectors and outputs the secondary localization result, i.e., the system-level localization label, such as "power system," "hydraulic system," "mechanical system," or "electrical system." The secondary classifier is trained on a fault sample set, which consists of fault sample sets from various systems. The label for each sample data is the system in which the fault occurred.

[0113] A three-level classifier is a classifier for component-level localization, such as three-level classifier III. i This is a classifier for the component-level localization of the i-th system. The input detection result is the feature vector of the fault, and the output is the component-level localization label of the i-th system, such as "engine" or "fuel tank" in a power system. (Three-level classifier III) i Training is performed on the fault sample set corresponding to system i. The training sample set consists of fault sample sets of each component. Each sample has the same system-level location label and the component-level location label is the component where the fault occurs.

[0114] In this embodiment, the individual classifiers in the first-level classifier, second-level classifier, and third-level classifier can employ classification algorithms including, but not limited to, decision trees, support vector machines, logistic regression, ridge regression, k-nearest neighbors, random forests, rotation forests, and neural networks.

[0115] Through the training process described above for each level of classifiers in the hierarchical localization model, a trained first-level classifier for fault detection, a second-level classifier for system-level localization, and a third-level classifier for component-level localization can be obtained, thus forming the hierarchical localization model. Using the trained classifiers at each level of the hierarchical localization model and employing a tree-structured fault localization process, label prediction can be performed on on-site equipment operation data to achieve the function of hierarchical fault localization.

[0116] The multivariate time series in the unlabeled device operation data are sequentially arranged and combined and transformed by random convolution kernels to generate feature vectors of different views, thus constructing a feature vector set. In this embodiment, the multivariate time series can be either historical series or the latest time series acquired in real time.

[0117] The feature vector set is input into the trained hierarchical localization model. The fault location at each level is predicted by the classifiers at each level according to the tree-like fault localization process. This is called hierarchical fault localization. Specifically, the feature vector set is input into the trained first-level classifier, and the first-level localization result is output. If the output first-level localization result is "normal", then the localization results at each level of the feature vector set are all "normal". If the output first-level localization result is "fault", then the fault detection result of the feature vector set is "fault", and the next level of fault localization needs to be performed.

[0118] The feature vector set indicating "fault" is input into the trained secondary classifier, and the output is the secondary localization result. In this embodiment, the secondary localization result is the predicted "system i". The feature vector of "system i" in the secondary localization result is input into the corresponding tertiary classifier III. i The output yields a three-level positioning result. In this embodiment, the three-level positioning result is the "component j" in the predicted system i.

[0119] Through the above-described tree-structured fault location process, the results of fault detection, system-level location, and component-level location of on-site equipment operation data can be predicted, and the location of faults at each level can be obtained, thereby realizing the function of hierarchical fault location.

[0120] To verify the accuracy of the fault location method provided in this embodiment, a fault location experiment was conducted on a validation set consisting of 100 labeled multivariate time series samples to verify the accuracy of the fault location method provided in this embodiment. Each time series sample in the validation set has a dimension of 4 and a length of 160 hours. The validation set includes 50 normal samples and 50 fault samples, and all fault samples are labeled with system-level and component-level location tags.

[0121] This embodiment performs five 5-fold cross-validation accuracy experiments based on the above validation set, with the average accuracy of each cross-validation being [missing value]. Figure 9 It will be displayed in the middle. Figure 9 The five dots represent the average accuracy of five fault detection results (i.e., first-level positioning results), the five X-shaped dots represent the average accuracy of five system-level positioning results (i.e., second-level positioning results), and the five square dots represent the average accuracy of five component-level positioning results (i.e., third-level positioning results).

[0122] pass Figure 9 The data shown clearly demonstrates that the fault location method provided in this embodiment can achieve high-precision fault location.

[0123] The fault location device provided by the present invention is described below. The fault location device described below and the fault location method described above can be referred to in correspondence.

[0124] Figure 10 This invention illustrates a fault location device provided in an embodiment of the present invention, the device comprising:

[0125] The acquisition module 1010 is used to acquire the device operation data of the target object; wherein, the device operation data includes a multivariate time series composed of various device operation parameters;

[0126] The first processing module 1020 is used to extract features from multivariate time series data to obtain a feature vector set.

[0127] The second processing module 1030 is used to perform hierarchical fault location on the feature vector set to obtain the fault location result.

[0128] In an exemplary embodiment, the first processing module 1020 described above can specifically be used for:

[0129] By arranging and combining the time series in a multivariate time series, multiple subsets of the series can be obtained;

[0130] Feature extraction is performed on each subset of sequences to obtain multiple feature vectors;

[0131] A feature vector set is constructed based on multiple feature vectors.

[0132] Furthermore, the first processing module 1020 described above can specifically perform feature extraction on each sequence subset through the following process to obtain multiple feature vectors:

[0133] Each sequence subset is transformed by a random convolution kernel to generate multiple feature maps;

[0134] A predetermined number of feature values ​​are extracted from each feature map to obtain the feature values ​​corresponding to each sequence subset.

[0135] Based on the feature values ​​corresponding to each sequence subset, feature vectors corresponding to each sequence subset are constructed to obtain multiple feature vectors.

[0136] In an exemplary embodiment, the second processing module 1030 described above can specifically be used for:

[0137] The feature vector set is input into a pre-built hierarchical localization model to obtain the fault localization result;

[0138] Among them, the hierarchical localization model is obtained by training a multi-level classifier based on feature vector set samples and the corresponding fault labels at each level of the feature vector set samples.

[0139] In an exemplary embodiment, the hierarchical localization model may include a first-level classifier, a second-level classifier, and a third-level classifier;

[0140] The second processing module 1030 mentioned above can be specifically used for:

[0141] The feature vector set is input into the first-level classifier to obtain the first-level localization result corresponding to the feature vector set;

[0142] If the first-level localization result of the feature vector set is a fault, then the feature vector set is input into the second-level classifier to obtain the second-level localization result corresponding to the feature vector set;

[0143] Based on the secondary localization results, the feature vector set is input into the corresponding tertiary classifier to obtain the tertiary localization results corresponding to the feature vector set;

[0144] The first-level, second-level, and third-level positioning results are used as the fault location results.

[0145] Furthermore, a first-level classifier can include multiple individual classifiers;

[0146] The second processing module 1030 described above can specifically achieve the following process: inputting the feature vector set into the first-level classifier to obtain the first-level localization result corresponding to the feature vector set:

[0147] Each feature vector in the feature vector set is input into the corresponding individual classifier to obtain the prediction results of each individual classifier.

[0148] Based on the prediction results of each individual classifier, the primary localization result is determined.

[0149] In an exemplary embodiment, the acquisition module 1010 described above can be specifically used for:

[0150] Obtain the raw data of the target object;

[0151] If the original data is found to contain missing values, these missing values ​​are filled in to obtain the device operation data.

[0152] In summary, the fault location device provided in this embodiment of the invention generates a sequence subset by arranging and combining multivariate time series data through a first processing module. Then, it extracts features from each sequence subset using a random convolution kernel transformation to generate a feature vector set composed of multiple feature vectors. This improves the diversity of features in each feature vector while considering the correlation between various equipment operating parameters. The second processing module uses a hierarchical location model and a tree-like fault location process to achieve serial fault detection, system-level location, and component-level location of the feature vector set, realizing the function of hierarchical fault location. Furthermore, the tree-like fault location process has good scalability and high location accuracy, effectively improving the fault location accuracy and reliability of the entire fault location device.

[0153] In addition, embodiments of the present invention also provide a working machine that uses the above-described fault location method or includes the above-described fault location device. By applying the above-described fault location method or fault location device, the working machine can achieve a more accurate and reliable fault location function, thereby improving the operational safety of the working machine.

[0154] In this embodiment, the operating machinery may include engineering operating machinery such as excavators and cranes.

[0155] Figure 11 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 11 As shown, the electronic device may include a processor 1110, a communications interface 1120, a memory 1130, and a communication bus 1140. The processor 1110, communications interface 1120, and memory 1130 communicate with each other via the communication bus 1140. The processor 1110 can call logical instructions in the memory 1130 to execute a fault location method. This method includes: acquiring device operating data of the target object; wherein the device operating data includes a multivariate time series composed of various device operating parameters; extracting features from the multivariate time series to obtain a feature vector set; and performing hierarchical fault location on the feature vector set to obtain a fault location result.

[0156] Furthermore, the logical instructions in the aforementioned memory 1130 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0157] On the other hand, the present invention also provides a computer program product, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, and when the program instructions are executed by a computer, the computer is able to execute the fault location method provided in the above embodiments, the method including: acquiring equipment operation data of a target object; wherein, the equipment operation data includes a multivariate time series composed of multiple equipment operation parameters; performing feature extraction on the multivariate time series to obtain a feature vector set; and performing hierarchical fault location on the feature vector set to obtain a fault location result.

[0158] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the fault location method provided in the above embodiments. The method includes: acquiring equipment operation data of a target object; wherein the equipment operation data includes a multivariate time series composed of various equipment operation parameters; extracting features from the multivariate time series to obtain a feature vector set; and performing hierarchical fault location on the feature vector set to obtain a fault location result.

[0159] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0160] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0161] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A fault location method, characterized in that, include: Acquire the equipment operation data of the target object; wherein, the equipment operation data includes a multivariate time series composed of various equipment operation parameters; the multivariate time series represents the change data of various equipment operation parameters over time; Feature extraction is performed on the multivariate time series to obtain a feature vector set; The feature vector set is used to perform hierarchical fault location to obtain fault location results; The feature extraction process for the multivariate time series, resulting in a feature vector set, includes: The time series in the multivariate time series are arranged and combined according to the type of equipment operation data to obtain multiple sequence subsets; Feature extraction is performed on each of the aforementioned sequence subsets to obtain multiple feature vectors; Based on multiple feature vectors, the feature vector set is constructed. The step involves extracting features from each of the sequence subsets to obtain multiple feature vectors, including: Each of the aforementioned sequence subsets is transformed by a random convolution kernel to generate multiple feature maps; A predetermined number of feature values ​​are extracted from each of the feature maps to obtain the feature values ​​corresponding to each of the sequence subsets; the predetermined number of feature values ​​includes the maximum value and the proportion of positive values ​​in the feature maps; Based on the feature values ​​corresponding to each of the sequence subsets, feature vectors corresponding to each of the sequence subsets are constructed respectively, resulting in multiple feature vectors; The step of performing hierarchical fault location on the feature vector set to obtain fault location results includes: The feature vector set is input into a pre-constructed hierarchical localization model to obtain the fault localization result; The hierarchical localization model is obtained by training a multi-level classifier based on feature vector set samples and the corresponding fault labels at each level of the feature vector set samples. The hierarchical localization model includes a first-level classifier, a second-level classifier, and a third-level classifier; The step of inputting the feature vector set into a pre-constructed hierarchical localization model to obtain fault localization results includes: The feature vector set is input into a first-level classifier to obtain the first-level localization result corresponding to the feature vector set; If the primary localization result is a fault, then the feature vector set is input into the secondary classifier to obtain the secondary localization result corresponding to the feature vector set; Based on the secondary localization result, the feature vector set is input into the corresponding tertiary classifier to obtain the tertiary localization result corresponding to the feature vector set; The first-level positioning result, the second-level positioning result, and the third-level positioning result are used as the fault location result; The primary classifier includes multiple individual classifiers; The step of inputting the feature vector set into a first-level classifier to obtain the first-level localization result corresponding to the feature vector set includes: Each feature vector in the feature vector set is input into the corresponding individual classifier to obtain the prediction results of each individual classifier. The primary localization result is determined based on the prediction results of each individual classifier; The acquisition of the target object's device operation data includes: Obtain the raw data of the target object; If it is determined that there are missing values ​​in the original data, the missing values ​​are filled in to obtain the device operation data.

2. A fault location device based on the fault location method according to claim 1, characterized in that, include: The acquisition module is used to acquire the device operation data of the target object; wherein, the device operation data includes a multivariate time series composed of various device operation parameters; The first processing module is used to extract features from the multivariate time series to obtain a feature vector set; The second processing module is used to perform hierarchical fault location on the feature vector set to obtain fault location results.

3. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the fault location method as described in claim 1.

4. A type of operating machinery, characterized in that, The operating machinery uses the fault location method as described in claim 1 or includes the fault location device as described in claim 2.

Images