JP5963579B2 - Excavator management device and excavator abnormality detection method - Google Patents

Description

  The present invention relates to an excavator management device and an excavator abnormality detection method.

  Work machines such as excavators are used in various construction sites, civil engineering sites, and the like, and when a failure occurs, quick failure repair is required. Various diagnostic systems have been developed that identify failure locations based on various operation information collected from work machines (Patent Document 1). However, when a failure occurs, the work must be interrupted until the failure repair is completed.

  An abnormality diagnosis technique for measuring the state of a work machine with a plurality of sensors and monitoring whether the work machine is normal has been developed (Patent Document 2). By detecting an abnormality before stopping the failure with the abnormality diagnosis technology and taking early maintenance measures, it is possible to prevent a failure of the work machine.

JP 2011-074639 A JP 2010-236302 A

The work machine is used in various environments such as extremely hot regions, cold regions, soft landfills, and hard mines. Moreover, it may be used continuously for 24 hours a day or only for a short time. Thus, it is difficult to accurately determine whether there is an abnormality in the shovel operating in various environments, usage patterns, etc. (operating conditions).
An object of the present invention is to provide an excavator management device and an excavator abnormality detection method capable of improving the detection accuracy of occurrence of abnormality.

According to one aspect of the invention,
An input device for inputting detected values of a plurality of variables depending on operating conditions from an excavator;
A processing device for processing detected values of the plurality of variables input from the input device;
The processor is
A set of detected values of the plurality of variables collected from the evaluation target excavator is used as an evaluation target sample,
The detected value of the plurality of set of variables and a one sample, from a plurality of samples are input from the input device, based on a detected value of said variable, operational conditions similar to operating conditions of the evaluation sample, And extract some samples collected during normal operation to make a standard sample group,
The evaluation sample, the excavator control apparatus be evaluated based on the standard sample group is provided.
According to another aspect of the invention,
Collecting the detected values of a plurality of variables depending on the operating state of the shovel or the surrounding environment as an evaluation target sample from the excavator;
From a plurality of samples each including detection values of the plurality of variables , based on the detection values of the variables , when the operating conditions are similar to the operating conditions of the sample to be evaluated and the excavator is operating normally Extracting a part of the collected samples into a standard sample group ,
There is provided an excavator abnormality detection method including a step of evaluating the sample to be evaluated based on the standard sample group.

  An abnormality can be detected more accurately by evaluating the evaluation target sample based on a standard sample group obtained by extracting some samples having similar detection values of variables.

FIG. 1 is a block diagram of an excavator management device according to the first embodiment. FIG. 2 is a list of samples collected from the shovel by the shovel management apparatus according to the first embodiment. FIG. 3 is a diagram illustrating data used in the abnormality detection method executed by the excavator management device according to the first embodiment and a processing flow of the abnormality detection method. FIG. 4A is an example of a distribution diagram of a sample for showing a method of classifying samples with similar degrees of detected values of the variable A and the variable B. FIG. 4B is a detailed data format of the variable C representing the mechanical load. FIG. 4C is an example of a sample distribution diagram for illustrating a method of classifying samples according to the degree of similarity of the detected value of the variable C. FIG. 5A is a sample list showing the results of classifying a plurality of samples, and FIG. 5B is a chart showing the results of classifying the evaluation target samples. FIG. 6 is a histogram of the variable k for explaining the effect of classifying a plurality of samples. FIG. 7 is a flowchart illustrating an excavator abnormality detection method according to the second embodiment. FIG. 8 is a diagram illustrating the data used in the excavator abnormality detection method according to the third embodiment and the processing flow of the abnormality detection method. FIG. 9 is a diagram illustrating a sample data structure and a sample data structure including only a variable of interest. FIG. 10A is a list of samples after extracting a target variable from a plurality of samples and a plurality of variables, and FIG. 10B is a data structure of an evaluation target sample after extracting a target variable from a plurality of variables. FIG. FIG. 11 is a diagram illustrating a flow of data and processing used in a method of selecting a target variable from a plurality of variables. FIG. 12 is a list of sample groups generated by associating sample groups and failure information with machine number and date information. It is a figure which shows a response | compatibility with the some variable which comprises a sample, and the gene of the chromosome used by a genetic algorithm. FIG. 14 is a flowchart of a genetic algorithm used in the excavator abnormality detection method according to the third embodiment. It is a figure which shows the flow of the process which evaluates a chromosome with a genetic algorithm, and the data used. FIG. 16 is a diagram illustrating the data used in the excavator abnormality detection method according to the fourth embodiment and the process flow of the abnormality detection method. FIG. 17 is a chart showing variables of interest selected by the method according to the fourth embodiment. FIG. 18 is a graph showing the frequency with which each variable is selected in the target variable list for each individual shovel.

[Example 1]
FIG. 1 shows a schematic block diagram of an excavator management apparatus according to the first embodiment. The excavator management device 20 includes a processing device 21, a failure information storage device 22, a sample data storage device 23, an input device 24, and an output device 25. Each of the plurality of excavators 26 measures a plurality of items (variables) depending on the operating state or operating conditions of the surrounding environment. Detection values of various variables measured by the excavator 26 are input to the input device 24 via the telecommunication line 28. The input device 24 may be a keyboard operated by an operator, an input interface for reading data from a removable storage device, or the like.

  The processing device 21 stores detection values of various variables input from the input device 24 in the sample data storage device 23 and performs an excavator abnormality detection process. Information on past failure cases is stored in the failure information storage device 22. The information on the failure case includes a machine number that identifies an individual shovel, date information on the occurrence of the failure, and details of the failure. The processing device 21 outputs the abnormality detection result to the output device 25. As the output device 25, for example, an image display device, a sound generation device, or the like is used.

  FIG. 2 shows an example of a sample collected from an excavator. Each of the samples includes the excavator machine number, date information obtained from the detected value, and detected values of a plurality of variables. Variables that depend on operating conditions include, for example, variables that depend on the environment and variables that depend on the driving situation. Examples of the environment-dependent variables include atmospheric pressure (variable A), intake air temperature (variable B), and the like. Examples of the variable depending on the driving situation include a mechanical load (variable C), a travel time ratio (variable D), and the like. FIG. 2 shows that the detected values of the variables A to G collected from the excavator of the machine number 1 on April 1, 2012 are x (1,1) to x (1,7), respectively. ing. The number of variables is generally 100 or more.

  FIG. 3 shows a processing flow of the excavator abnormality detection method according to the first embodiment and data used. In step ST1, the processing device 21 (FIG. 1) collects samples from the shovel and stores them in the sample data storage device 23 (FIG. 1). Thereby, the sample group 30 consisting of a plurality of samples is prepared in the excavator management device 20. In step ST2, the plurality of samples are classified into a plurality of sample groups based on the degree of similarity of the detected values of the variables.

  An example of a method for classifying a plurality of samples will be described with reference to FIG. 4A. FIG. 4A shows a distribution diagram of a sample focusing on variable A (atmospheric pressure) and variable B (intake air temperature). Samples distributed in regions with low atmospheric pressure suggest working at high altitudes, for example. Samples distributed in regions where the intake air temperature is low suggest, for example, work in cold regions. Thus, when the work environment is different, the detected value of the variable may be different depending on the work environment. For example, samples having an atmospheric pressure of less than 50 kPa and an intake air temperature of less than 0 ° C. are classified into the first sample group 31. Samples having an intake air temperature of 50 ° C. or higher are classified into the third sample group 33. Samples not included in any of the first and second sample groups 31 and 33 are classified into the second sample group 32. The environment in which the samples classified into the first sample group 31 are collected is high altitude and low temperature and severe. The environment in which the samples classified into the third sample group 33 are collected is a severe environment having an air temperature of 50 ° C. or higher.

Another example of the classification method will be described with reference to FIGS. 4B and 4C. FIG. 4B shows an example of the data format of the variable C indicating the machine load. The range from 0% to 100% of the mechanical load factor is divided into 10 stages, and the time t1, t2,... Operated at the mechanical load factor of the relevant category is stored for each category. For example, adding the times t8, t9, and t10 operated at a mechanical load factor of 70 to 80%, 80 to 90%, and 90 to 100%, the time t (C ≧ 70) operated at a mechanical load factor of 70% or more is calculated. can do.
FIG. 4C shows a distribution diagram of samples focusing on the variable C (mechanical load). The horizontal axis of FIG. 4C represents the time during which the machine load factor is 70% or more, and the vertical axis represents the frequency (number of samples). In the example shown in FIG. 4C, it is shown that the frequency of operating for 2 to 3 hours with the mechanical load factor of 70% or more is the highest. For example, a sample operating at a mechanical load factor of 70% or more, a sample of 5 hours or more, a sample of 1 to 5 hours, and a sample of less than 1 hour are designated as the first sample group 31 and second sample group 32 and the third sample group 33. The operating conditions in which the samples classified into the first sample group 31 are collected are operating conditions with severe mechanical loads such as excavation of hard rock.

  In FIG. 4A, samples are classified by focusing on variable A and variable B. In FIG. 4C, samples are classified by focusing on variable C, but focusing on variable D (traveling time ratio) shown in FIG. The sample may be classified, or the sample may be classified by focusing on other variables.

  As shown in FIG. 3, the plurality of samples included in the sample group 30 are classified into one of the first sample group 31, the second sample group 32, and the third sample group 33. FIG. 5A shows a list of samples after a plurality of samples are classified. In the example shown in FIG. 5A, samples with sample numbers 1, 5, and 7 are classified into the first sample group 31, samples with sample number 4 are classified into the second sample group 32, and sample numbers 2, 3, and The 8, 9, and 10 samples are classified into the third sample group 33.

  In step ST3 shown in FIG. 3, the evaluation object sample 35 is collected from the evaluation object shovel. FIG. 5B shows the data structure of the evaluation target sample 35. The detected values of the variable A to the variable G of the evaluation target sample 35 are y (1) to y (7). In step ST4, a sample group to which the evaluation target sample 35 belongs is determined. In the example shown in FIGS. 3 and 5B, the evaluation target sample 35 belongs to the first sample group 31.

  In step ST5, only samples collected during a period in which the excavator is operating normally are extracted from the plurality of samples of the first sample group 31 to which the evaluation target sample 35 belongs. A sample group including only samples collected during a period of normal operation is referred to as a standard sample group 34.

  In step ST6, the evaluation target sample 35 is evaluated using the standard sample group 34 as a reference. Various multivariate analysis methods can be applied as the evaluation method. By this evaluation, it is evaluated whether or not the evaluation object sample 35 is abnormal in the standard sample group 34, that is, whether or not an abnormality has occurred in the evaluation object shovel. The evaluation result is output from the output device 25.

  Note that, before classification in step ST2, samples at the time of failure may be removed from the entire sample group 30, and only samples during normal operation may be targeted for classification. In this case, the first sample group 31 includes only samples during normal operation. For this reason, step ST5 can be omitted and the first sample group 31 can be used as the standard sample group.

  FIG. 6 shows an example of a distribution diagram (histogram) of samples focusing on the variable k. 6 indicates the distribution of a plurality of samples belonging to the standard sample group 34, and the broken line 30 indicates the distribution of a plurality of samples belonging to the entire sample group 30. The detected value of the variable k of the evaluation target sample 35 is y (k). When the distribution of the entire sample group 30 is used as a criterion for determination, the detected value y (k) of the variable k is considered to be a general value rather than an abnormal value. However, when the distribution of the standard sample group 34 is used as a criterion for determination, it can be seen that the detected value y (k) of the variable k is an abnormal value greatly deviating from the value indicating the maximum frequency.

  As shown in FIG. 6, when the entire sample group 30 is used as an evaluation criterion when evaluating the evaluation target sample 35, the detected value y (k) of the variable k of the evaluation target sample 35 is abnormal. I miss it. In Example 1, the entire sample group 30 is classified into a plurality of sample groups 31 to 33 in advance. As the evaluation target sample 35, a standard sample group 34 similar to the environment or operating condition in which the evaluation target sample 35 is measured is used as an evaluation criterion. For this reason, it is possible to reduce oversight of abnormality.

[Example 2]
With reference to FIG. 7, a description will be given of an excavator abnormality detection method according to the second embodiment. In the second embodiment, the evaluation method in step ST6 shown in FIG. 3 is embodied.

  In step SA1, based on the standard sample group 34 (FIG. 3), a unit space used in the Mahalanobis Taguchi method is generated. In step SA2, the Mahalanobis distance MD from the center of the unit space to the evaluation target sample 35 (FIG. 3) is calculated. When calculating the Mahalanobis distance MD, all the variables (FIG. 2) constituting the sample may be used, or only some of the variables may be used.

  In step SA3, it is compared whether or not the calculated Mahalanobis distance is greater than a threshold value. If the Mahalanobis distance MD is less than or equal to the threshold value, it is determined in step SA4 that the evaluation target sample 35 is normal. If the Mahalanobis distance MD is greater than the threshold value, it is determined in step SA5 that the evaluation target sample 35 is abnormal.

  As in Example 2, it is effective to use the Mahalanobis Taguchi method as a method for evaluating whether the evaluation target sample 35 is normal or abnormal.

[Example 3]
The excavator abnormality detection method according to the third embodiment will be described with reference to FIGS. In Example 1, in step ST6 shown in FIG. 3, the evaluation was performed while paying attention to all variables A, B, C,... (FIG. 2) constituting the sample. In Example 3, the evaluation is performed using only some of the plurality of variables.

  As shown in FIG. 8, a sample group 30, failure information 40, and an evaluation target sample 35 are prepared. The sample group 30 and the evaluation target sample 35 are the same as the sample group 30 and the evaluation target sample 35 (FIG. 3) handled in the first embodiment. Similar to the method of the first embodiment shown in FIG. 3, the standard sample group 34 is generated from the sample group 30. The failure information 40 includes past failure cases that have occurred in the shovel. Each failure case includes an excavator machine number, date information on which the failure occurred, and a failure content.

  In step ST10, some variables are selected from a plurality of variables constituting each sample included in the sample group 30. The selected variable is referred to as a target variable 50.

  FIG. 9 shows an example of the selected variable of interest 50. Each sample of the sample group 30 (FIG. 8) includes a machine number, date information, and detected values of variables A, B, C,. As an example, variables E, H, J, L, M, P,... Are excluded from a plurality of variables, and variables A, B, C, D, F, G, I, K, N, O are included. It is selected as the variable of interest 50.

  FIG. 10A and FIG. 10B show a list of standard sample groups 34 in which the target variable 50 is specified, and an example of the evaluation target sample 35. As shown in FIG. 10A, the plurality of samples in the standard sample group 34 are classified into a plurality of sample groups by the same technique as that classified in step ST2 (FIG. 3) of the first embodiment. The standard sample group 34 includes samples of sample numbers 1, 5, 7,... Classified into the first sample group 31 (FIG. 3). As the variable of interest 50, variables A, B, C, D, F, G, I, K, O,. As shown in FIG. 10B, also in the evaluation target sample 35, variables A, B, C, D, F, G, I, K, O,.

In step ST12, the evaluation target sample 35 is evaluated using the standard sample group 34 as an evaluation criterion and using only the variable of interest 50. This evaluation is the same as the evaluation method of step ST6 (FIG. 3) adopted in Example 1 or the evaluation method (FIG. 7) adopted in Example 2. After the evaluation, in step ST13, the evaluation result is output to the output device 25 (FIG. 1).

  With reference to FIGS. 11 to 15, a method of selecting some target variables from a plurality of variables in step ST <b> 10 (FIG. 8) will be described.

  As shown in FIG. 11, in step ST <b> 20, the sample group 30 and the failure information 40 are associated with each other by the machine number and date information and combined. As a result, a sample group 41 including the failure content included in the failure information as a variable is obtained.

  FIG. 12 shows a list of sample groups including failure contents as variables. Each sample includes a machine number, date information, variables A, B, C,. The failure content variable indicates the content of the failure that occurred on the date specified by the date information in the excavator specified by the machine number variable of each sample. A sample in which “0” is set in the variable of the failure content indicates that no failure has occurred.

  In step ST21 of FIG. 11, a plurality of samples of the sample group 41 including the failure content as a variable is classified into any of the first to third failure content-containing sample groups 42 to 44. The classification method is the same as the method of step ST2 (FIG. 3) used in the first embodiment.

  In step ST22, a sample group to which the evaluation target sample 35 (FIG. 8) belongs is determined. This determination method is the same as the method of step ST4 (FIG. 3) used in the first embodiment. FIG. 11 shows an example in which the evaluation target sample 35 belongs to the first failure information-containing sample group 42.

  In step ST23, the samples of the first failure information-containing sample group 42 whose failure content is set to “0” are classified into the normal sample group 45, and some failure is set as the failure content. Those are classified into a sample group 46 at the time of failure. A variable of interest is selected by applying a genetic algorithm based on the normal sample group 45 and the fault sample group 46.

  FIG. 13 shows a gene 61 of a chromosome (individual) 60 used in the genetic algorithm. The plurality of genes 61 correspond to a plurality of variables constituting samples included in the sample group 30 (FIG. 11). A gene can take two values, “adopted” and “not adopted”. For example, “1” and “0” are defined as “adopted” and “non-adopted”, respectively.

  FIG. 14 shows a flowchart of the genetic algorithm used in the third embodiment. In step SB1, an initial solution (initial generation) of the chromosome 60 is generated. The value of each gene of the plurality of chromosomes 60 in the initial generation may be set at random, or the number of chromosomes 60 for which “adopted” is set for the gene 61 corresponding to the variable that should be adopted as the variable of interest. May be more. In step SB2, each of the chromosomes 60 is evaluated.

  A detailed procedure of step SB2 will be described with reference to FIG. In step SB21, one chromosome 60 is selected from a plurality of chromosomes as an evaluation target. A unit space is generated based on samples included in the normal sample group 45 using only variables corresponding to genes having a value of “adopted” in the chromosome 60 to be evaluated.

In step SB22, the Mahalanobis distance 48 from the center of the unit space 47 is calculated for each of the plurality of failure samples of the failure sample group 46. In step SB23, the Mahalanobis distance 48 of each sample at the time of failure and the threshold 49 of the Mahalanobis distance.
And compare. Based on the number of samples in failure having a Mahalanobis distance exceeding the threshold value 49, the fitness of the chromosome 60 is determined.

  The large number of failure samples having a Mahalanobis distance exceeding the threshold value 49 means that the failure samples can be easily distinguished from the normal samples. Therefore, the higher the fitness, the higher the number of failure samples having Mahalanobis distances exceeding the threshold 49.

  In step SB3 shown in FIG. 14, a chromosome having a high fitness is selected from the plurality of chromosomes 60. As a selection method, methods such as elite selection and roulette selection may be applied. In step SB4, a new chromosome is generated by causing crossover or mutation in the selected plurality of chromosomes 60. In step SB5, it is determined whether an end condition is satisfied. As an example, the end condition is satisfied when the generation of chromosomes reaches a predetermined generation.

  If the termination condition is not satisfied, the chromosome generation is updated in step SB6. Specifically, the chromosome selected in step SB3 is left, and the chromosome not selected is deleted. Further, the new chromosome generated by crossover or mutation in step SB4 is added as a new generation chromosome. The procedure from Step SB2 to Step SB5 is repeated based on the new generation of chromosomes.

  If it is determined in step SB5 that the end condition is satisfied, in step SB7, a combination of variables of interest selected from a plurality of variables is determined based on the chromosome of the last generation. As an example, a variable corresponding to a gene for which a value of “adopted” is set is selected as a variable of interest in the chromosome with the highest fitness among the chromosomes of the last generation.

  In Example 3, since a variable that makes it easy to determine that the sample at the time of failure is abnormal is used as the variable of interest, the abnormality of the evaluation target sample 35 can be more accurately determined as abnormal.

[Example 4]
The excavator abnormality detection method according to the fourth embodiment will be described with reference to FIGS. Hereinafter, differences from the third embodiment will be described, and description of the same configuration will be omitted. In Example 3, the variable of interest 50 was selected as one sample group without distinguishing the excavator machine number in step ST10 shown in FIG. For this reason, as shown in FIG. 10A, the same variable of interest 50 is selected from a plurality of variables depending on operating conditions, regardless of the excavator machine number. In Example 4, a variable of interest is selected for each excavator machine number (individual).

  FIG. 16 shows the relationship between the data for selecting the variable of interest from a plurality of variables depending on the operating conditions. In selecting the variable of interest, the data of the sample group 30 and the failure information 40 are used as in the third embodiment shown in FIG. Each sample stored in the sample group 30 includes information for identifying the excavator machine number (individual) as shown in FIG. Therefore, a plurality of samples can be classified for each machine number, and a sample list for each machine number can be generated. Similarly, a failure information list for each machine number can be generated from the failure information 40.

In step ST30, a variable of interest is selected for each excavator machine number from a plurality of variables A, B, C, D,... Depending on operating conditions (FIG. 2). As a method for selecting the target variable, for example, the same method as the method of the third embodiment described with reference to FIGS. 11 to 15 can be applied. However, in the case of the fourth embodiment, in the process of selecting the target variable of the machine number i (i = 1, 2, 3,...), The sample of the machine number i out of the plurality of samples of the sample group 30. Only the list is used, and only the failure information list of the machine number i is used among the plurality of pieces of failure information stored in the failure information 40. Accordingly, the target variable list 50A for machine number 1, the target variable list 50B for machine number 2, the target variable list 50C for machine number 3, and so on are selected. The target variable list 50A for machine number 1 and the target variable list 50B for machine number 2 are not necessarily the same.

  FIG. 17 shows an example of the variable of interest selected in step ST30 (FIG. 16) in a table format. In FIG. 17, variables that are not selected as the variable of interest are shaded, and variables that are selected as the variable of interest are outlined. In the example shown in FIG. 17, variables A, B, C, D, F, G, I, K, O, and the like are included in the target variable list 50 </ b> A of machine number 1, and variables A, C, D, E, I, J, M, O, P, etc. are included in the target variable list 50B of machine number 2, and variables A, B, C, D, F, G, J, K, L, etc. are the target of machine number 3. It is included in the variable list 50C. In this way, when the excavator machine number is different, the selected variable of interest is also different.

  In step ST31 (FIG. 16), the common target variable 51 is extracted from the target variable lists 50A, 50B, 50C,. Next, a method for extracting the common target variable 51 in step ST31 will be described with reference to FIG. For each of a plurality of variables depending on operating conditions, the frequency of selection in the target variable list 50A, 50B, 50C,.

  FIG. 18 shows the frequency with which each variable is selected in the target variable list 50A, 50B, 50C,. The horizontal axis represents a variable depending on the operating condition, and the vertical axis represents the frequency (number of times) selected as the variable of interest. The variables on the horizontal axis are rearranged in the order of the selected frequency. Based on the frequency selected as the variable of interest, some of the variables of interest are adopted as the common variables of interest from the plurality of variables of interest. As an example, a variable with a high frequency selected as a target variable is preferentially adopted as a common target variable. FIG. 18 shows a case where a plurality of (for example, 50) variables are employed as the common attention variable 51 in order from the variable with the highest frequency.

  In step ST32 (FIG. 16), the evaluation target sample 35 is evaluated using an evaluation model using the common sample variable 51 with the standard sample group 34 as an evaluation criterion. The standard sample group 34 and the evaluation target sample 35 are the same as the standard sample group 34 and the evaluation target sample 35 (FIG. 8) used in Example 3, respectively. The process of step ST32 is the same as the process of step ST12 (FIG. 8) of the third embodiment.

  In step ST33, the evaluation result performed in step ST32 is output. The process of step ST33 is the same as the process of step ST13 (FIG. 8) of the third embodiment.

  In the fourth embodiment, the target variable that has a small influence on the determination result is excluded from the evaluation model in advance. The evaluation target sample is evaluated with an evaluation model using a common target variable that is considered to have a large influence on the evaluation result. For this reason, highly accurate evaluation can be performed with fewer target variables.

  Although the present invention has been described with reference to the embodiments, the present invention is not limited thereto. It will be apparent to those skilled in the art that various modifications, improvements, combinations, and the like can be made.

20 excavator management device 21 processing device 22 failure information storage device 23 sample data storage device 24 input device 25 output device 26 excavator 28 telecommunications line 30 sample group 31 first sample group 32 second sample group 33 third sample group 34 Standard sample group 35 Evaluation target sample 40 Failure information 41 Sample group 42 including failure information as a variable First failure content-containing sample group 43 Second failure content-containing sample group 44 Third failure content-containing sample group 45 Normal Sample group 46 Failure sample group 47 Unit space 48 Mahalanobis distance 49 Mahalanobis distance threshold 50 Target variable 50A, 50B, 50C Target variable list 51 for each machine number Common target variable

Claims (10)

An input device for inputting detected values of a plurality of variables depending on operating conditions from an excavator;
A processing device for processing detected values of the plurality of variables input from the input device;
The processor is
A set of detected values of the plurality of variables collected from the evaluation target excavator is used as an evaluation target sample,
The detected value of the plurality of set of variables and a one sample, from a plurality of samples are input from the input device, based on a detected value of said variable, operational conditions similar to operating conditions of the evaluation sample, And extract some samples collected during normal operation to make a standard sample group,
The evaluation sample, the management apparatus shovel be evaluated based on the standard sample group. The processor is
Select some of the plurality of variables as the target variable,
The excavator management device according to claim 1, wherein when the evaluation target sample is evaluated based on the standard sample group, the evaluation target sample is evaluated by an evaluation model using the variable of interest. The processor is
3. A failure sample consisting of detected values of the plurality of variables collected at the time of a shovel failure is evaluated by the evaluation model, and the variable of interest to be selected from the plurality of variables is determined based on an evaluation result. The excavator management device described in 1. The plurality of variables include a machine number for identifying an individual shovel and date information,
Furthermore, it has a failure information storage device in which the machine number of the shovel where the failure occurred, the date information of the failure occurrence, and the failure information including the failure content is stored,
The processing device associates the plurality of samples input from the input device with the failure information stored in the failure information storage device based on the machine number and the date information, thereby The excavator management device according to claim 3, wherein a sample is generated. The processor is
The combination of the variables of interest is determined by repeatedly evaluating the sample at the time of failure using the evaluation model while changing the combination of the variables of interest selected from the plurality of variables. The excavator management device described. In the processing in which the processing device repeatedly evaluates the sample at the time of the failure,
A plurality of chromosome genes are respectively associated with the plurality of variables,
The value of each gene is represented by one of two values corresponding to “adopted” and “non-adopted” of the variable corresponding to the gene,
Generate an initial solution containing multiple chromosomes with different combinations of gene values,
Until the termination condition is satisfied, the chromosome generation is updated by performing evaluation, selection, crossover, and mutation on the chromosome,
The excavator management device according to claim 5, wherein a combination of the variables of interest is determined based on a chromosome of the last generation. Collecting the detected values of a plurality of variables depending on the operating state of the shovel or the surrounding environment as an evaluation target sample from the excavator;
From a plurality of samples each including detection values of the plurality of variables , based on the detection values of the variables , when the operating conditions are similar to the operating conditions of the sample to be evaluated and the excavator is operating normally Extracting a part of the collected samples into a standard sample group ,
A method for detecting an abnormality of a shovel comprising a step of evaluating the sample to be evaluated based on the standard sample group. Furthermore, it has a step of selecting a part of variables from the plurality of variables to be a target variable,
The excavator abnormality detection method according to claim 7, wherein when the evaluation target sample is evaluated based on the standard sample group, the evaluation target sample is evaluated by an evaluation model using the variable of interest. further,
A step of evaluating a failure sample consisting of detected values of the plurality of variables collected at the time of a shovel failure with the evaluation model;
The excavator abnormality detection method according to claim 8, further comprising: determining the target variable to be selected from the plurality of variables based on an evaluation result of the failure sample. The plurality of variables include a machine number for identifying an individual shovel and date information,
Further, by associating the machine number of the excavator in which the failure occurred, the failure occurrence date information, the failure information including the failure content, and a plurality of samples collected from the excavator based on the machine number and the date information. The method for detecting an abnormality of the excavator according to claim 9, further comprising a step of generating the failure sample.

Images