JP6329892B2 - Construction machine management system - Google Patents

Description

  The present invention relates to a construction machine management system that diagnoses the state of a construction machine and its components using oil test results.

  Construction machinery employs a hydraulic system that rotates at high speed and high load, such as a hydraulic pump and a hydraulic motor. Further, there is lubricating oil in the gears of the traveling device and the turning device. In construction machinery, the quality of the lubricating oil such as hydraulic fluid, engine oil, and gear oil has a great influence on mechanical performance and reliability such as power transmission efficiency and damage prevention. Therefore, a method has been considered in which the oil used in the construction machine is analyzed to diagnose the state of the oil and the machine, thereby preventing failure and performance deterioration.

  Japanese Patent Publication No. 4-68580 (Patent Document 1) is known as a conventional failure diagnosis system. In this publication, as a main purpose, it is described that the determination of the failure diagnosis system that can automatically determine the diagnosis of the internal state of the machine to be diagnosed based on the oil analysis result is provided (second publication of the publication). (Right column, lines 11 to 14). As a method for realizing this, rank evaluation is performed based on a predetermined reference value for each metal element or mixed foreign matter, and oil analysis values are combined in time series with rank-analyzed oil analysis values. Form a matrix, create a failure pattern that shows the correlation between rank evaluation combination and failure occurrence, read a combination with a high failure rate from this failure pattern to form a failure matrix pattern, oil analysis matrix and failure matrix pattern And determining whether there is a match or not. Further, it is described that when there is a match, the determination result or the maintenance measure corresponding to the determination result is displayed on the display device (the first column, the lower left column, claim 1).

Japanese Examined Patent Publication No. 4-68580

  According to the technology of Patent Document 1, the oil analysis matrix obtained by determining the rank of the oil test result with a preset management reference value is compared with the failure matrix pattern called from the file, and the type of the failure matrix that matches. (Failure code)-The risk level and maintenance measures set in advance according to the number are displayed on the display device (Fig. 3 in the publication).

  However, since the contents displayed on the display device are only the “risk level” and “maintenance measures”, the logical configuration leading to the diagnosis result from the test result cannot be understood.

  If the user is a maintenance worker who performs maintenance work, understanding the logical configuration can improve the quality of the maintenance work, and performing maintenance work without understanding the logical structure means that the quality of the maintenance work May be reduced. Even if maintenance work is not performed directly, and maintenance work is outsourced, if the logical configuration can not be understood, the maintenance time (time for ordering maintenance work) and the contents are adjusted. I can't do that either.

  An object of the present invention is to provide not only a diagnosis result but also a logical configuration that has led to the diagnosis result to a user, thereby enabling appropriate construction machine maintenance work.

  The present invention includes a plurality of means for solving the above-mentioned problems. For example, the management criteria for the test items of oil collected from construction machines, diagnosis rules, and diagnosis results are combined to make a final diagnosis. A recording unit that records a logical configuration, an oil test result acquisition unit that acquires a test result of oil collected from a construction machine, an evaluation value is obtained from the test result, and based on the evaluation value and the management criteria Based on at least one of the evaluation value and the determination result, the diagnosis rule and the logical configuration, a state diagnosis of the construction machine or its components is performed based on the evaluation value. A diagnostic processing unit that generates an explanatory document or a logical configuration diagram of the diagnostic result based on the diagnostic rule and the logical configuration leading to a diagnostic result is provided.

  According to the present invention, the user can grasp the diagnosis process from the oil test result to the diagnosis result being derived together with the diagnosis result.

It is a block diagram of the management system of a construction machine. It is the figure which showed the example of the flow of the whole process of an oil diagnostic system. It is the figure which showed the list of oil test items. It is the figure which showed the flowchart of the determination process of an oil test result. It is the figure which showed the example which sets a management reference value with an upper limit, a new oil value, and a magnification. It is the figure which showed the example of the management reference value with respect to oil use time. It is the figure which showed the example which calculates | requires the linear model of the management reference value with respect to oil use time. It is the figure which showed the example of the relationship between the management reference values with respect to oil use time. It is the figure which showed the example of distribution of the iron content data in an oil test. It is the figure which showed the example of distribution of the iron content data after double logarithmic transformation. It is the figure which showed the example which applied generation | occurrence | production of iron content data to double logarithmic normal distribution. It is the figure which showed the example of distribution of the silicon content data in an oil test. It is the figure which showed the example which applied generation | occurrence | production of silicon | silicone content data to double logarithmic normal distribution and double logarithmic Weibull distribution. It is the figure which showed the example applied to distribution of iron content, and double logarithmic normal distribution. It is the figure which showed the data structure of a diagnostic rule, a diagnostic result, and a determination result. It is the figure which showed the flowchart of the diagnostic process. It is the figure which showed the flowchart of the diagnostic rule application process in a diagnostic process. It is the figure which showed the example of the increase in the diagnostic result in a diagnostic process, and the end of a diagnostic process. It is the figure which showed the example of the diagnostic rule. It is a figure explaining the chain relationship of a diagnostic result and a diagnostic rule. It is the figure which showed the example of the chain relationship of a diagnostic result. It is the figure which showed the example of the relationship between the chain relationship of a diagnostic result, and a tree structure. It is the figure which showed the example which circulates a tree structure and produces | generates a text. It is a figure explaining the phrase and the output in the classification | category of each element of a tree structure. It is the figure which showed the flowchart of the document production | generation process of a diagnostic result. It is the figure which showed the example of the format of a report. It is the figure which showed the example of the format of the comment column of a diagnostic result. It is the figure which showed the example of the recommendation rule. It is the figure which showed the flowchart of the recommendation process. It is the figure which showed the flow of the data between a customer, a construction machine maker / agent, and an oil analysis specialized company.

  FIG. 1 is a configuration diagram of a construction machine management system. The construction machine management system of this embodiment includes oil test apparatuses 1 to n (reference numerals 101 to 103), various DBs 121 to 125, a display apparatus 180, and a computer (computer) 114 connected to these via a network as software. An oil diagnosis system 131 that realizes oil diagnosis processing in cooperation is provided.

  The oil test apparatuses 101 and 103 include computers (computers) 111 and 113 and are connected to a network. The test results of the oil test apparatuses 101 and 103 are directly stored in the test result DB 121 via the network. On the other hand, when a computer is not installed like the oil test apparatus 102, the test result is input via a network by inputting the test result with an input device (not shown) of the computer 112 connected to the network. It is stored in the test result DB 121, or by being input with an input device (not shown) of the calculator 114, it is stored in the test result DB 121.

  In addition to the test result DB 121, the database includes a new oil information DB 122 for storing information on test values of new oil meaning unused oil, a management standard DB 123 for storing management standards for determining oil test results, and diagnostic rules. A diagnosis rule DB 124 for storing information and a recommended rule DB 125 for storing recommended information for associating recommended information such as actions recommended to customers and agents and diagnosis results are provided.

  The oil diagnosis system 131 includes an oil test result acquisition unit 141 that acquires oil test result data from the test result DB 121, and can acquire an oil test result that is necessary when executing each process. Further, as shown in FIG. 1, the oil diagnosis system 131 includes (1) a plurality of processing units 151 to 154 related to the management reference setting process, and (2) a plurality of processing units 161 to 165 related to the oil test result determination process, (3) A plurality of processing units 171 to 179 related to the diagnostic processing and a plurality of processing units classified into three are provided. Hereinafter, functions of these processing units will be described.

  In the management standard setting process, a management standard setting unit 151 that sets a threshold for each rank, that is, a management standard value, necessary for determining the rank of the oil test result for each test item is used. Test results related to metal wear powder and external contaminants may be set as fixed threshold values for each rank, but the test results related to properties and additives will be a certain test value (new oil value) for new oil. The test value varies depending on use. For this reason, you may set a management reference | standard by the relative value (For example, the difference with respect to a new oil value or the magnification with respect to a new oil value, etc.) with respect to a new oil value.

  Oil deteriorates according to the usage time of construction machinery. Therefore, the oil usage time correction setting unit 152 performs a setting for correcting the management reference value with respect to the oil usage time. If the actual use time of the construction machine cannot be measured accurately, the oil test value has a correlation between the elapsed time from the oil change to the time of oil collection and the actual use time of the construction machine. The elapsed time may be used as the usage time.

  In addition, metal wear powder and external contaminants that cause deterioration of oil are generated differently depending on the model and equipment. Therefore, first, the oil test result acquisition unit 141 acquires past oil test result data of a specific model and device. Next, the double logarithmic transformation statistical modeling unit 153 obtains a distribution of test value occurrence probabilities from past oil test result data. Then, the statistical management reference value setting unit 154 obtains a test value corresponding to the probability of occurrence of abnormality, and sets this as a management reference value. This setting is obtained in advance and stored as a management reference value.

  More specifically, the double logarithmic transformation statistical modeling unit 153 acquires the test value of the oil test result that has been performed in the past and accumulated in the test result DB 121 by the oil test result acquisition unit 141, and performs the test. Probabilistic model of occurrence of test value after double logarithmic conversion by calculating the average and standard deviation of test values after double logarithmic conversion. As a normal distribution, a double log normal distribution is calculated. Then, the statistical management reference value setting unit 154 calculates a test value after double logarithmic conversion for a preset cumulative probability value (probability of occurrence of abnormality), and calculates the test value after double logarithmic conversion. Is subjected to double exponential conversion (that is, exponential conversion is performed after exponential conversion), a test value corresponding to the cumulative probability value is calculated, and a test value corresponding to the calculated cumulative probability value is set as a management reference value .

  Note that the test value after double logarithmic conversion may not be applied to the normal distribution. In that case, the probability of occurrence is modeled by a Weibull distribution. In this case, the double logarithmic transformation statistical modeling unit 153 acquires the test value of the oil test result that has been performed in the past and accumulated in the test result DB 121 by the oil test result acquisition unit 141, and sets the test value to 2 In order to obtain a Weibull distribution (that is, a double logarithmic Weibull distribution), the test value after double logarithmic conversion is arranged in ascending order in order to obtain a rank. By dividing the ranking by the total number of data, the cumulative probability value of occurrence of the test value after double logarithmic conversion is calculated, the value obtained by logarithmically converting the test value after double logarithmic conversion is an explanatory variable, and the cumulative probability of occurrence Using a value obtained by converting the value based on the Weibull cumulative distribution formula as an objective variable, the coefficient and intercept of the linear expression are calculated by the least square method, and the coefficient of the linear expression is used as the shape parameter of the Weibull distribution. Weibull from coefficient and intercept To calculate the scale parameter. Based on the double logarithmic Weibull distribution modeled by the double logarithmic transformation statistical modeling unit 153 as described above, the statistical management reference value setting unit 154 performs a double logarithmic transformation on a preset cumulative probability value. Test value is calculated, test value after double logarithm conversion is converted to double exponential, test value corresponding to cumulative probability value is calculated, and test value corresponding to calculated cumulative probability value is managed Set as reference value.

  The setting of the management reference value by the double logarithmic conversion statistical modeling unit 153 and the statistical management reference value setting unit 154 will be described in detail later using specific examples of FIGS. 9 to 14.

  By the way, in the oil test result determination process, first, the oil test result acquisition unit 141 acquires the oil test result to be determined. A management standard is acquired from the management standard DB 123 by the management standard acquisition unit 161. When the test item is a property or additive, since the management standard is set with reference to the new oil value, the new oil information acquisition unit 162 acquires new oil information from the new oil information DB 122. The determination threshold setting unit 163 obtains a management reference value corresponding to the numerical value of the test value in the range of the determination rank. If the management standard is set as a numerical value that directly corresponds to the test value, the determination threshold setting unit 163 may use the management standard value for directly determining the value acquired by the management standard acquisition unit 161. . Then, the determination unit 164 ranks the oil test results. The determination result is output to the display device by the determination result output unit 165. However, the determination result is stored in the memory of the computer 114 or the hard disk in order to use the determination result in the diagnosis process, or stored together with the test result in the test result DB 121. Good.

  In the diagnosis process, first, a diagnosis rule is acquired by the diagnosis rule acquisition unit 171 from the diagnosis rule DB 124. The diagnosis rule of this embodiment is defined in the IF-THEN format, and expresses a one-layer logical relationship that results in the THEN part if the IF part condition is satisfied. A plurality of diagnosis rules are created in advance based on past diagnosis results and know-how, and all of them are stored in the diagnosis rule DB 124. One or more determination results and / or one or more diagnosis results are set in the IF part of each diagnosis rule, and one diagnosis result is set in the THEN part. That is, the diagnosis rule can output not only a new diagnosis result from the determination result but also a new diagnosis result from the diagnosis result.

  Next, in the diagnosis process, the determination result acquisition unit 172 acquires one or more determination results related to the oil test result determination process. Then, all the diagnosis rules acquired by the diagnosis rule acquisition unit 171 from the diagnosis rule DB 124 are applied to all the determination results acquired by the determination result acquisition unit 172 one by one. When the condition set in the IF part of the diagnosis rule applied by the diagnosis part 173 is satisfied, a diagnosis result of the THEN part of the diagnosis rule is newly obtained. At that time, the diagnostic linkage relation acquisition unit 174 links the newly obtained diagnostic result and the condition (determination result, diagnostic result) set in the IF part of the diagnostic rule with the diagnostic rule. Then, the diagnosis unit 173 applies all the diagnosis rules one by one again to all the determination results acquired by the determination result acquisition unit 172 and the new diagnosis result, and when a new diagnosis result is obtained, The diagnostic linkage relation acquisition unit 174 links the IF part and the THEN part of the diagnostic rule that generated the new diagnostic result. Thereafter, this is repeated until no new diagnosis result is obtained. Through these processes, one or more determination results and / or one or more determination results set in the IF part of each diagnosis rule that has generated a diagnosis result from when the first diagnosis rule is applied until no new diagnosis result is obtained. The diagnosis result and the diagnosis result set in the THEN unit are linked to obtain a chain structure in which the determination result and the diagnosis result are nodes and the diagnosis rule is a directed link (chain). The obtained chain structure indicates a logical structure from the determination result acquired by the determination result acquisition unit 172 to the final diagnosis result (final diagnosis result). The obtained chain structure is usually one or more directed tree structures, and the diagnosis result located at the end of each structure is the final diagnosis result.

  Note that the diagnosis result set in the THEN part (result) of each diagnosis rule is limited to one, and each diagnosis rule does not generate the same diagnosis result (the same diagnosis result from the diagnosis rule in the diagnosis rule DB 124). Are not generated). The chain structure is a directed tree without a loop. This indicates that, when a certain diagnosis result is used as a starting point and the chain of diagnosis results linked by the diagnosis rule is successively followed, the original diagnosis result is not restored. Therefore, the final diagnosis result always exists. There may be multiple final diagnostic results. By tracing the edge leading to each final diagnosis result in the reverse direction via the diagnosis rule, the logical structure from which the final diagnosis result is derived from the determination result is specified.

  The logical structure leading to the final diagnosis result thus obtained is documented for reporting to a person such as a construction machine user or an agent. Here, a sentence in which the logical structure leading to the final diagnosis result is documented may be referred to as a diagnosis sentence.

  Specifically, first, from the individual chain structure (linkage relationship) indicating the logical structure leading to the final diagnosis result acquired by the diagnosis chain relationship acquisition unit 174, the diagnosis tree structure acquisition unit 175 generates a tree structure (referred to as a diagnosis tree structure). Get it). The diagnosis tree structure can be obtained by removing the diagnosis rule from the chain structure. In the tree structure, the final diagnosis result is the root node, the oil test result determination result is the leaf node, the final diagnosis result (root node) and the determination result ( The diagnosis result located between the leaf nodes becomes an internal node. The acquired tree structure is converted into a sentence by the diagnostic sentence generation unit 176. The diagnosis tree structure has a logical relationship from the leaf node to the root node via the internal node. Therefore, it is possible to create a logical sentence by traversing the tree structure.

  Specifically, the recommended rule acquisition unit 177 first acquires a recommended rule associated with the final diagnosis result from the recommended rule DB 125. Next, the recommendation unit 178 creates a recommended sentence that is a sentence in which an action that a user, a construction machine manufacturer, or an agent desirably performs is described based on the acquired recommendation rule and the final diagnosis result.

  The determination result of the oil test result, the diagnostic text, and the recommended text are output to the display device 180 as a report in a predetermined format by the report output unit 179. Note that the output method is not limited to display on the display device, but is notified to the user, the agent, or the like in the form of data (recording medium or network) provision or paper provision.

  In addition, if each hardware which comprises the said system is comprised so that communication is mutually possible, there will be no limitation in particular in the installation place of each hardware. Furthermore, the functions of each computer may be integrated to reduce the number of computers, or conversely, the functions of each computer may be distributed to increase the number of computers.

  The computer 114 is connected to a display device (display) 180, a printer, and other computers or servers via a network, and can exchange various data. This data includes oil diagnostic reports.

  According to the present embodiment configured as described above, the threshold value (management reference value) used for determining the test result of the oil test item (ranking the degree of abnormality) is optimally set. The degree can be determined appropriately. A final diagnosis result can be derived by repeatedly applying a plurality of prepared diagnosis rules to the combination of determination results of each test item. In addition, based on the IF part and THEN part of the diagnostic rule that has generated a new diagnostic result until the final diagnostic result is acquired, the logical structure from the determination result to the final diagnostic result can be easily acquired, Using this, the logical structure can be expressed as a sentence or a tree structure. That is, according to the present embodiment, the determination of the degree of abnormality of the test result that has been performed by an oil analysis expert so far, the final diagnosis based on the determination result, and the determination result has led to the final diagnosis. The logical structure can be easily grasped. Furthermore, since this embodiment uses a recommendation rule that associates the final diagnosis result with the recommended text, the report containing the determination result, the diagnostic text, and the recommended text can be automatically obtained simply by obtaining the oil test result. Can be created. In addition, when the logical structure from the determination result to the final diagnosis result is expressed in text, a person other than an oil analysis expert can easily interpret the content of the failure that has occurred in the construction machine.

  In addition, a management reference value can be calculated from actual data by using a statistical calculation model by double logarithmic transformation. In particular, for test items belonging to the classification of metal wear powder and external contaminants, management reference values should be set according to the type, amount generated in actual operation, and the detected value in the test for each device. Thus, the management reference value can be automatically set more appropriately.

  Hereinafter, a more detailed specific example in the case of diagnosing the state of the construction machine and its equipment using the above construction machine management system will be described.

<1. Oil analysis>
First, oil analysis will be described. A user of a construction machine collects oil of each device mounted on the construction machine using a sampling kit for collecting oil. The equipment is a structure to be lubricated that is mounted on a construction machine, such as an engine, various gears such as a transmission, a travel speed reducer, a turning speed reducer, or a hydraulic system. In each device, engine oil, gear oil, and hydraulic oil are used for lubrication. Oil samples taken with the kit are sent to an oil analysis company where various oil tests are conducted. Normally, oil testing is conducted by an oil analysis company that is independent of the construction machine manufacturer, but it may also be a construction machine manufacturer or a company that has a capital relationship with it.

  Various tests are the pentane that evaluates the degree of acidity (neutralization number) such as kinematic viscosity, base number / total acid number, flash point, fuel content, water content, and solid soil content in oil. Testing oil properties such as insoluble matter, ICP (Inductive-Coupled Plasma) plasma emission spectroscopy (SOAP) to detect elements such as iron and silicon in oil, and measuring the amount of molecules in oil This is a test such as Fourier Transform InRrared (FT-IR) spectroscopy. Elements that can be detected by plasma emission spectroscopic analysis are (1) metal wear powder generated by wear of sliding parts of equipment such as iron, copper and lead, and (2) silicon and sodium from outside the machine or lubrication structure. (3) Additives added to improve oil performance such as phosphorus, zinc and calcium.

  From the oil test results, the condition of the oil and the condition of the equipment are diagnosed. The diagnosis result is notified as a report to the construction machine user and to the construction machine manufacturer and its agent. The agency provides the user with an appropriate service according to the diagnosis result.

  FIG. 2 shows the overall processing flow of the oil test result determination, diagnosis, and report generation. In FIG. 2, characters surrounded by an ellipse indicate processing, characters not surrounded by an ellipse indicate data, arrows indicate a data flow, and the rest indicates a database. In addition, the same code | symbol is attached | subjected to the same part as the previous figure (it is the same also in later figures).

  First, the management reference setting 201 is performed before the determination 214 of the oil test result. The management standard setting 201 is manually set by a person, and the computer 114 executes other rounding processes. The management standard is stored in the management standard DB 123. Metal wear powder and external contaminants are generated differently depending on the type and equipment of the construction machine. Therefore, the statistical setting 204 of the management reference value is executed based on the past oil test result data acquired from the oil test result DB 121, and the management reference value is stored in the management reference DB 123.

  The oil test result determination process 214 is executed with the oil test result 211, the oil use information 212, the management standard of the management standard DB 123, and the new oil information of the new oil information DB 122 as inputs. Oil usage information 212 includes information such as customer and machine serial information, oil sampling information including the date of oil collection, model and equipment information, oil brand and viscosity grade information, oil usage time, and after the start of machine operation. It is the machine operating time meaning the total operating time. In the oil test result determination process 214, the rank determination process is performed by referring to the management standard of the management standard DB 213 and, if necessary, the new oil information DB 122 and the oil use information and comparing them with the input oil test result 211. And the determination result 215 is obtained.

  When the determination result 215 is obtained, the diagnosis process 222 is executed using this as an input. In the diagnosis processing 222, a new diagnosis result is obtained by applying the diagnosis rule acquired from the diagnosis rule DB 124 to the determination result 215. When a new diagnosis result is obtained, the diagnosis rule of the diagnosis rule DB 124 is applied again to the result obtained by adding the determination result 215, and thereafter, until a new diagnosis result is not obtained, the final diagnosis result is obtained. 223 is obtained. Further, the determination result 215 and the diagnosis result are appropriately linked based on the IF part and the THEN part of the diagnosis rule that generated the diagnosis result during the diagnosis process 222, and the determination result 215 linked with the diagnosis rule is diagnosed. The resulting chain structure is obtained.

  In the sentence generation process 224, a sentence indicating a logical structure from the determination result 215 to the diagnosis result 223 is generated based on the chain structure obtained in the diagnosis process 222. On the other hand, in the recommendation process 232, the recommended rule acquired from the recommended rule DB 125 is applied to the diagnosis result 223 to obtain a recommended sentence.

  The diagnosis sentence and the recommended sentence obtained in the sentence generation process 224 and the recommendation process 232 are described in the report 241 together with the determination result 215. Although an arrow is not drawn in FIG. 2, the oil test result 211 and the oil use information 212 that are inputs to the system are described in the report 241 as information that is a premise of the diagnosis process 222.

<2. Method of judging oil test results>
An oil test result determination method will be described. FIG. 3 shows a list of oil test items, and FIG. 4 shows a flowchart of determination processing.

  First, oil test items and management criteria for determination will be described. As shown in FIG. 3, the oil test items are classified into four categories: properties, metal wear powder, external contaminants, and additives. Since the test item represents the state of oil or machine, it is sometimes called a state index.

  In the present embodiment, all test results are quantified (numerized), and here, the numerical values are used as test values. However, there may be a case where a predetermined conversion is required for the test value when compared with the control standard. Therefore, in the present embodiment, whether the test value is used as it is or when a value obtained by converting the test value is used, the value is referred to as an “evaluation value”. Therefore, the determination of the test result is to determine the degree of abnormality of the evaluation value of the test result. In this embodiment, an example is described in which a numerical range of test values is set for each test item, and determination is performed by rank. However, the number of ranks does not need to be uniform for each test item, and the number of ranks varies depending on the test item. May be allowed.

  Hereinafter, a case will be described in which the rank number 4 is set, specifically, ranks indicating four levels of Normal (normal), Alert (caution), Urgent (emergency), and Critical (fatal) are set. For example, when the iron management standard value is the upper limit value of 40 ppm for Normal, the upper limit value of 70 ppm for Alert, the upper limit value of 100 ppm for Urgent, and the larger range is Critical, the determination result is 50 ppm when the test value is 50 ppm. It becomes Alert.

  The test value of iron belonging to the classification of metal wear powder is limited to a value of 0 or more, and the determination is limited to a range of 0 or more. When the range of the test value is limited to a certain value or more, the determination direction is “upper”. When the test value is limited to a certain value or less, the determination direction is “lower”, and when a test value is generated above or below a certain value, the determination direction is “both sides”. is there.

  Test values of test items belonging to properties and additive classifications need to be determined based on test values (new oil values) of unused new oil, depending on the oil brand and viscosity grade. For test items that depend on the oil brand and viscosity grade, a new oil value is required to set the control standard value. Additives are set at the new oil value only for the added elements. In the example of FIG. 3, new oil values are required because P, Zn, and Ca are added to the oil, while no new oil values are required because B, Ba, Mg, and Mo are not added.

  In the above iron example (in which the control standard value is the upper limit value of 40 ppm for Normal, the upper limit value of 70 ppm for Alert, the upper limit value for Urgent is 100 ppm, and the larger range is Critical), the test result corresponds to any rank. The test value itself was determined as an evaluation value.

On the other hand, when determining based on the new oil value, the difference between the new oil value and the test value is determined as the evaluation value. The direction of determination is the upper side, the lower side, or both sides. If the direction of determination is classified, the absolute value of the difference between the new oil value and the test value may be determined as the evaluation value. The determination content is shown by taking a kinematic viscosity of 40 ° C as an example. The new oil value is 100 mm ² / s. The upper limit value for the management reference value is set to ± 15%. In other words, the upper limit value is 115 mm ² / s and the lower limit value is 85 mm ² / s. However, in order to evaluate the absolute value, the management reference value is determined based on the difference between the upper limit value and the new oil value. More specifically, the management reference value is set by a magnification that is multiplied by the upper limit value of the difference between the upper limit value and the new oil value. If the upper limit value of Normal is 0.5 times, the upper limit value of Alert is 0.8 times, and the upper limit value of Urgent is 1 time, the actual management standard when the difference between the upper limit value and the new oil value is used as the evaluation value The values are Normal upper limit value 0.5 × 15 = 7.5 mm ² / s, and the same calculation results in Alert upper limit value 12 mm ² / s and Urgent upper limit value 15 mm ² / s. For example, if the test value was 110 mm ² / s, the evaluation value is 110-100 = 10mm ² / s, and the so this is comprised between 7.5~12mm ² / s, the rank decision is Alert .

<2-1. Management standards>
Here, the management reference value setting method and the management reference for setting the management reference value will be described. First, regarding the management standard for determining the test value itself such as iron, a fixed value may be set for each rank such as Normal and Alert as the management standard value. The fixed value set as the upper limit value and the lower limit value of the test value for each rank is the management reference value.

  As a management standard for determining the test value itself, a management standard value may be set according to the magnification of the new oil value. For example, when a certain amount of the additive is added to the new oil, and there is an increase or decrease, it can be diagnosed that “the additive has been consumed” or “an oil of a different brand has entered”. Therefore, the upper value may be set to Normal for up to 2 times, Alert for up to 3 times, Urgent for up to 4 times, and Critical for more than the new oil value. On the lower side, Normal is set from 1 to 0.7 times, Alert is set to 0.4 times, Urgent is set to 0.2 times, and Critical is set below that. If the new oil value of phosphorus (P) is 2000 ppm, the upper limit is set to a normal upper limit value of 4000 ppm, the Alert upper limit value of 6000 ppm, the Urgent upper limit value of 8000 ppm, and the lower side is set to the Normal lower limit value of 1400 ppm, the Alert lower limit value of 800 ppm, and the Urgent lower limit value of 400 ppm. Thereby, the test value itself can be determined. The new oil value and its magnification are the management criteria.

  In the above description, the new oil value and the magnification thereof are used as the management standard based on the new oil value. However, any numerical value may be set as the reference value instead of the new oil value. In this case, the reference value and its magnification become the management reference.

  When the absolute value of the difference between the new oil value and the test value is used as the evaluation value, the management reference value can be set as a fixed value. For example, the new oil value of phosphorus (P) is 2000 ppm, and the management reference value of the absolute value of the difference between the new oil value and the test value is set to the upper limit value of 2000 ppm, the Alert upper limit value of 4000 ppm, and the Urgent upper limit value of 6000 ppm. . In this case, if the test value is 3200 ppm, the evaluation value is | 2000-3200 | = 1200 ppm, and since it is less than the normal upper limit value, it is determined as normal.

  Even when the absolute value of the difference between the new oil value and the test value is used as the evaluation value, it is possible to set the management reference value according to the magnification. For this purpose, an upper limit value is required for the upper side, and a lower limit value is required for the lower side. In the case of the upper side, the threshold value is obtained by setting the difference between the upper limit value and the new oil value to 1 and multiplying by the magnification for each rank. FIG. 5 shows an example in which the management reference value is set by the scale factor for the difference between the upper limit value 502 and the new oil value 501. The upper limit value of Normal was 0.5 times (512), the upper limit value of Alert was 1 time (513), and the upper limit value of Urgent was 2 times (514).

  As an example of a specific setting in this case, there is the setting of the kinematic viscosity of 40 ° C. described above. The upper limit was set to + 15% with respect to the new oil value. For each rank, the normal upper limit value is 0.5 times, the Alert upper limit value is 0.8 times, the Urgent upper limit value is 1 time, the upper limit value of the new oil value is obtained, and the difference between the upper limit value and the new oil value is calculated. The management standard value was set by multiplying each rank.

  It may be necessary to correct the control reference value for the oil usage time. This is because the oil deteriorates as the device is used, and the metal wear powder may increase according to the amount of contact at the sliding portion, that is, the sliding distance. Therefore, a correction for increasing the control reference value with respect to the oil usage time is modeled on the assumption that the test value increases with respect to the oil usage time. FIG. 6 shows the relationship of the management reference value to the oil usage time. The vertical axis is the test value, and the horizontal axis is the oil usage time. First, it is assumed that management standards for each rank are set as management reference values 611, 612, and 613 in the figure. A time point at which the management reference value is to be managed is defined as a reference oil usage time 601. It is assumed that the correction model is a straight line. In order to determine a straight line, it is only necessary to determine two points through which the straight line passes. Therefore, the management reference values 621, 622, and 623 at the time when the oil use time is 0 hours (at the time of oil change) may be set. The management reference value at 0 hour is its initial value, and setting the initial value of the management reference value is setting an intercept as a linear model. The slope of the straight line model is the amount of change in the management reference value per unit time.

  FIG. 7 shows an intercept setting method. First, the intercept value α 701 can be simply set. Next, the intercept value α 701 can also be determined by setting a difference amount β 702 with respect to the management reference value γ 703 in the reference oil usage time. Alternatively, the intercept value α 701 can be determined by setting the ratio of the intercept to the control reference value γ 703, that is, the value of α ÷ γ.

  The intercept of each rank may be set directly for the management reference value of each rank. If it is assumed that a certain relationship is established for the management reference value of each rank, the intercept of each rank can be obtained from the intercept relationship in one rank by designating the relationship. As the relationship between the management reference values, there are a parallel relationship (shift) of the straight line model for correcting each management reference value, and a constant ratio relationship (ratio). FIG. 8 shows an intercept for the relationship. (A) The method for obtaining the intercept when the management reference value is parallel will be described. It is assumed that management reference values 801, 802, and 803 are set for the reference oil usage time, and an intercept b813 is set. In this case, the difference between the management reference value 801 and the intercept b813 is taken, and the intercept b811, b812 is obtained by subtracting the difference from the management reference values 801 and 802, respectively. (B) The method of obtaining the intercept in the case where the ratio is constant (ratio) will be described. The ratio is constant means that the ratio of the management reference value of each rank is equal to the ratio of the intercept, that is, α: α + β: α + β + γ = α ′: α ′ + β ′: α ′ + β ′ + γ ′ is established. It is. Assume that the intercept b833 is set. Therefore, the ratio = (α + β + γ) ÷ (α ′ + β ′ + γ ′) is determined. Each intercept is obtained as intercept b832 = ratio × (α ′ + β ′) and intercept b831 = ratio × α ′.

  In the test value determination method in which the management reference value is corrected with respect to the oil usage time, first, the management reference value corrected with respect to the oil usage time is obtained by the oil usage time of the tested sample. Specifically, when a model is determined by a straight line, an intercept may be added by multiplying the slope by the oil usage time. The slope is obtained by subtracting the intercept from the control reference value and dividing by the reference oil usage time. A determination result is obtained by determining the rank of the evaluation value with respect to the corrected management reference value.

<2-2. Judgment process>
The determination process will be described with reference to the flowchart of FIG. The following processing is executed by the computer 114. When there is no sentence subject, the subject is the computer 114. First, the oil information and the oil test result of the model to be determined and the device are acquired (step 401).

  Next, the management reference information of the construction machine and equipment to be determined is acquired (step 402). Also, new oil value data is acquired (step 403). The management reference value is not always set directly as in the iron content example, but as shown in the example of the kinematic viscosity of 40 ° C., the reference information such as the new oil value information and the magnification for setting the management reference value Is required. Therefore, the information necessary for setting the management reference value, including the direct management reference value, is called a management reference.

  The range from step 404 to step 415 is a loop process for all test items. In this one-time loop, first, the presence / absence of a test value is determined (step 405). If the test is not performed, if the test fails, or if the oil itself is not in the range of normal contamination etc. and the test result is abnormal, the test value is skipped and the judgment process for the test item is skipped. To do.

  Next, it is determined whether or not there is a management standard (step 406). Although management criteria are acquired in step 402, the management criteria are not necessarily registered in the database 123 for all test items. When there is no management standard, the determination process for the test item is skipped.

  Next, it is determined whether or not new oil data is necessary (step 407). For example, since the new oil data is unnecessary for iron, the determination result is negative. In the example of the kinematic viscosity of 40 ° C., since a new oil value is necessary, the presence / absence of new oil data is determined (step 408). When the new oil data is not registered in the database 122, the determination process for the test item is skipped.

  Then, the determination direction is determined (step 409), and the process proceeds to the determination process. The direction is divided into the upper side, the lower side, and both sides for judging both the upper and lower sides. Since the determination process itself is the same process flow in each direction, it is described in a single arrow in the flowchart.

  In the determination, first, an evaluation value (test value) is acquired or calculated (step 410). In addition, the management standard value of the standard oil usage time is set. The management reference value is a threshold value for each rank. When the management standard value is not directly set in the management standard, the management standard value is obtained using the management standard information and, if necessary, the new oil value (step 411).

  Next, it is determined whether or not the management reference value is corrected with respect to the oil usage time, that is, whether or not the time is corrected (step 412). If it is determined in step 412 that time correction is to be performed, oil usage time correction is set (step 413). This is to define a straight line which is a correction model.

  Then, the rank is determined based on the evaluation value and the state is set (step 414). After performing the loop processing from step 404 to step 415 for all the test items, the determination result is output (step 416).

<3. Method of setting management standards from past oil test result data>
A method for statistically setting management standards for metal wear powder and external contaminants from past oil test result data will be described. The metal wear powder includes substances depending on the material of the sliding portion, such as iron (Fe), lead (Pb), copper (Cu), chromium (Cr), aluminum (Al), and tin (Sn). External contaminants include substances such as silicon (Si), which is the main component of sand, and sodium (Na) added to the coolant (coolant). For the rank judgment of metal wear powder and external contaminants, the test value itself is used as the evaluation value, and the new oil value is not used. The amount of metal wear powder and external contaminants mixed in the oil, and the amount mixed per oil unit volume vary depending on the model and equipment. Therefore, a statistical occurrence distribution of past actual data is obtained for each model and device, and a management reference value is calculated for the occurrence probability.

The steps of the statistical management standard value setting method are as follows.
(1) Define management reference values for one specified model and device.
(2) The statistical generation distribution is obtained from the actual oil test result data of the model and equipment of (1). The occurrence probability with respect to the management reference value is obtained and set as the management reference probability value.
(3) A statistical distribution of occurrence is obtained from data on the actual oil test results of models and devices other than (1). The management reference value is obtained from the management reference probability value.

  First, the management reference value of one designated model and device is determined. This designated model is assumed to be model A. A device is a component of a machine such as an engine, a hydraulic system, and various gears. As a model, there are many shipments, and it is a representative model such as a popular model and a model that has been used for a long time. This is because knowledge has also been obtained about machine failures. In order to determine the management reference value of this model, the test value at that time may be used for a malfunction such as a past actual failure. When data on test values and the presence / absence of defects are accumulated over a long period of time, a test value that increases the probability of a defect may be used as a management reference value. Moreover, it is good also as the management reference value corresponding to the model and apparatus made into the object made public.

  Next, the statistical generation distribution of the test values is obtained from the data of the oil test result of the model A equipment. FIG. 9 shows (a) the distribution of data with respect to the oil usage time and (b) the frequency distribution of data with respect to the test values. From FIG. 9A, it can be seen that most of the data is 50 ppm or less, while some data exceeds 200 ppm. As shown in FIG. 9B, the maximum frequency is 12 ppm and gradually decreases on the right side. This distribution on the right side is a very wide distribution in which the frequency 0 times continues to a maximum value exceeding 200 ppm. This frequency distribution cannot be applied to known probability distributions such as normal distribution and exponential distribution. Therefore, the distribution is modeled by logarithmic transformation of the data.

  The result of logarithmically transforming the data twice is shown in FIG. This conversion is called double logarithmic conversion here.

  In the above formula (1), x is a test value. At the time of logarithmic conversion, the Napier number e is added to the test value. This is because the test value of metal wear powder and external contaminants is 0 ppm at the minimum, and becomes negative infinity when the logarithm is taken directly. In order to perform double logarithmic conversion, the Napier number e is added so that the converted value becomes 0 at 0 ppm. From FIG. 10, it can be seen that the data after the double logarithmic transformation has a distribution in which the skirt spreads uniformly from 0 to 2 with 1 as the center.

  The data after double logarithmic transformation is modeled as a normal distribution. This model is referred to as a double-log-normal distribution.

  In the above equation (2), z is a test value after the double logarithmic transformation of equation (1), μ is the average of the data, and σ is the standard deviation. FIG. 11A shows the fitting result. The horizontal axis represents the value after double logarithmic conversion, and the vertical axis represents the cumulative occurrence probability. The solid line is the estimated line by the model, and the broken line is the cumulative probability value of the actual data. For this data, the model fits well.

  FIG. 11B shows the cumulative probability with respect to the test value. Therefore, the occurrence probability for the management reference value 1101 is obtained and set as the management reference probability value 1102. It is to obtain a random variable value that becomes a cumulative probability value. The cumulative probability value of the double lognormal distribution is as follows.

  A management reference probability value is calculated by subjecting the management reference value to a double logarithmic transformation using Equation (1) and substituting it into Equation (3). If the management reference value is set to a normal upper limit value of 40 ppm, an Alert upper limit value of 70 ppm, and an Urgent upper limit value of 100 ppm, the corresponding management reference probability value is calculated as 92%, 98%, and 99% if the decimal point is rounded down. It was.

  Test values after double logarithmic transformation may not fit into the normal distribution. As an example, FIG. 12 shows the distribution of silicon data. About the magnitude | size of dispersion | variation, it turns out that it is distribution with a very wide bottom like FIG. However, as shown in FIG. 12B, the distribution is monotonically decreasing, and even if the data is subjected to double logarithmic transformation, it cannot be a bilaterally symmetric distribution like a normal distribution. Even if the distribution does not decrease monotonously, there is a possibility that it is not symmetrical with respect to the average value or the mode value like the normal distribution. Therefore, the occurrence probability is modeled by a Weibull distribution from which various forms of probability density distribution can be obtained. The Weibull distribution can be a distribution having different shapes according to the scale parameter λ and the shape parameter α. The probability density function of the Weibull distribution is Expression (4), and the cumulative distribution function is Expression (5).

  In the above formulas (4) and (5), z is the value after the double logarithmic transformation shown in formula (1). Therefore, the distribution of the equation (4) is called a double logarithmic Weibull distribution (double-log-Weibull distribution).

  The scale parameter and the shape parameter of the Weibull distribution can be calculated using data by the Weibull plot method. Equation (5) can be modified as follows.

  The cumulative probability value F (z) of data can be obtained by arranging the data in ascending order, taking the rank, and dividing by the number of data. Lnz is also a numerical value. Therefore, if the left side is set to Y, lnz = X, and −αlnλ = C, Expression (7) is obtained.

  The shape parameter α and the intercept C are obtained by the method of least squares. The scale parameter can be calculated by equation (8).

  FIG. 13 shows the result of fitting the data in FIG. 12 to (a) a double log normal distribution and (b) a double log Weibull distribution. As shown in FIG. 13A, when the double lognormal distribution is used as a model, this is not true for the actual data generation probability. As shown in FIG. 13B, when the double logarithmic Weibull distribution is used as a model, it can be understood that the model is well applied.

  The calculation of the management reference probability value is the same procedure as the double lognormal distribution. A management reference probability value is obtained by subjecting the management reference value to double logarithmic transformation and then substituting it into equation (5).

  Next, a method for setting management reference values for models other than model A will be described. One model other than model A is model B. The device is the same for model A and model B. First, a statistical distribution of occurrence is obtained from the data of the model B oil test result. FIG. 14A shows the iron data distribution of model B. FIG. A small number of data is several hundred ppm, indicating that the distribution is very wide. FIG. 14B shows the cumulative probability of occurrence with the horizontal axis as the test value, which is the result of fitting to the double lognormal distribution. Compared to FIG. 11B, it can be seen that model B generates a lot of data in the range of test values smaller than model A. For example, in FIG. 11B, the cumulative probability 0.9 is 40 ppm, but in FIG. 14B, the cumulative probability 0.9 is 20 ppm. Obtaining the management reference value from the statistical occurrence distribution is obtaining the management reference value 1402 from the management reference probability value 1401. The calculation process is a process for obtaining the value of z on the right side by using the value on the left side of Expression (3) or Expression (5) as the management reference probability value. Since both Equation (3) and Equation (5) are monotonically increasing functions, the value of z can be obtained by numerical analysis calculation such as Newton's method. For the formula (5), the management reference value of the test value x is calculated by the following calculation.

  If the management reference probability value is set to 92% of the upper limit value of Normal, 98% of the upper limit value of Alert, and 99% of the upper limit value of Urgent, the corresponding management reference value is calculated as 21 ppm, 37 ppm, 47 ppm by rounding down the decimal point. .

<4. Oil diagnosis method>
Next, an oil diagnosis method will be described. The diagnosis process is a process for obtaining a diagnosis result from the determination results of a plurality of test items using the result of the rank determination of the oil test result. The diagnosis result finally obtained has a complex logical structure in which a plurality of determination results and diagnosis results have a chain relationship.

  The types of diagnosis rules stored in the diagnosis rule DB 124 include: (1) When one or more determination results set in the condition part (described later) exist, the diagnosis results set in the result part (described later) A first type diagnostic rule that generates one and (2) a second type diagnostic rule that generates one diagnostic result set in the result part when there is one or more diagnostic results set in the condition part (3) When there is one or more determination results and one or more diagnosis results set in the condition part, 3 of the third type diagnosis rule that generates one diagnosis result set in the result part There is one. However, there is no diagnosis rule that generates the same diagnosis result. Diagnostic rules are set in advance based on past diagnosis results and expert know-how, taking into account all judgment result combination patterns, and all diagnostic rules necessary for diagnosis based on any judgment result combination Are stored in the diagnostic rule DB 124. Note that the number of each of the first type, second type, and third type diagnosis rules is zero or more, that is, there is no limitation on the number of each diagnosis rule.

  In the diagnosis process, all the diagnosis rules stored in the diagnosis rule DB 124 are sequentially applied to the determination result of the test item actually tested to obtain the diagnosis result. The result obtained by adding the initial determination result to the diagnosis result thus obtained becomes the application target of the next diagnosis rule. Therefore, all the diagnosis rules stored in the diagnosis rule DB 124 are applied again to these application targets, and further diagnosis results are obtained. Thereafter, this is repeated until no new diagnostic result is obtained even if all the diagnostic rules are applied.

  The diagnostic rule stored in the diagnostic rule DB 124 includes a condition part and a result part. One or more determination results and / or one or more diagnosis results are set in the condition part, and only one diagnosis result is set in the result part. When the determination result corresponding to the condition part and the diagnosis result have already been obtained, that is, when the condition part is satisfied, the content of the result part is set as a new diagnosis result. The diagnosis rule is formulated by equation (11). The diagnosis rule is an IF-THEN rule.

  Here, “∧” means logical product, “¬” means negation operator, and “/” means exclusion. “B / ¬B” means B or not B. The IF part is a condition part, and the THEN part is a result part. The condition part has one or more conditions, that is, n is 1 or more. If all conditions are satisfied, the result B / ¬B is obtained. Satisfying all the conditions of a certain diagnosis rule is said to “fire”.

  Satisfying the condition means that the obtained test item or diagnosis result and its state fall within the range of the test item or diagnosis result set as a condition and its state. For example, when the determination result that the iron content of the test item is Urgent has been obtained and the condition is that the iron content is in the range of more than Alert, that is, any of Alert, Urgent, Critical, the condition is Satisfied.

  A condition test item or a diagnosis result is called a condition item. FIG. 15 shows the data structure of one condition and result part, as well as diagnosis results and determination results. Since the test item is determined in the upper and lower ranges with respect to the new oil value or the like, the direction is set in the case of the test item. When setting the state range, the above state range is set to (NOT_LESS) and below (NOT_MORE). In the argument item, if the item is abnormal wear, additive increase, or additive decrease in the diagnosis result, the target substance is specified.

  A diagnosis result is set in the result part. If the diagnosis result is abnormal wear, additive increase, or additive decrease, specify the target substance in the reference item. Similarly to the test items, the diagnosis results are also provided with categories corresponding to the phenomena, and the property property, metal wear wear, external contamination Ingress, and additive additive are set in the diagnosis categories.

  Since both the diagnosis result and the determination result are applied to the diagnosis rule, they have the same data structure. Only the diagnosis result is obtained from the diagnosis rule, and the determination result is not obtained. What is peculiar to the determination result is that there is a direction of large / small and that the diagnostic classification is a test Test. The state is a result of rank determination in the case of a determination result, and in the case of a diagnosis result, a state with the highest degree of malfunction is set among test items and diagnosis results that satisfy the condition part.

  The diagnosis process will be described with reference to the flowcharts of FIGS. Since the processing is performed by the computer 114, when the subject of the sentence is not particularly described, the subject is the computer 114.

  First, the determination result of the model and device to be determined is acquired (step 1601). Also, the diagnosis rule for the model and device to be determined is acquired (step 1602). Then, the determination result is set as an initial diagnosis result list, and the list backup is emptied (step 1603). This corresponds to initialization of diagnostic processing.

  The basic idea of the process for deriving a complex logical structure was to first apply all diagnostic rules to the determination result (initial determination result) to obtain the diagnosis result, and then to obtain it All diagnostic rules are applied again to the diagnosis result and the initial determination result to obtain a new diagnosis result. This is repeated, and if no new diagnosis result is obtained, the final diagnosis result is output. It is considered that the processing has been completed. That is, as long as there is a diagnostic rule that fires, a logical connection is continuously generated using the fired diagnostic rule. Therefore, in this embodiment, the diagnosis rule is applied in an infinite loop, the diagnosis results obtained thereby are accumulated in a list (diagnosis result list), and a backup of the diagnosis result list (list backup) is performed for each loop. The method of updating and ending when the diagnosis result list matches the list backup list was used. The infinite loop here is to perform loop processing without setting the number of loops in advance. The infinite loop corresponds to the range from step 1604 to step 1615 in FIG. Let this infinite loop be loop 1.

  FIG. 18 shows an example of changes and termination of the diagnosis result list and list backup in an infinite loop. First, determination results 1, 2, and 3 are set in the diagnosis result list 1801 as an initial state. The list backup 1802 is empty. In the first loop, the contents of the diagnosis result list are first copied to the list backup. All diagnosis rules are applied to the determination results 1, 2, and 3 in the diagnosis result list, and the two diagnosis rules are fired to obtain new diagnosis results 1 and 2 (1803). When the loop ends, the contents of the diagnosis result list 1801 and the list backup 1802 are compared. At the end of the first loop, the contents of the diagnosis result list 1801 and the list backup 1802 do not match, so execution of the second loop is determined. Even in the second loop, the contents of the diagnosis result list are first copied to the list backup. All diagnosis rules are applied to the determination results 1 and 2 and the diagnosis results 1 and 2 in the diagnosis result list, and one diagnosis rule is fired to obtain a new diagnosis result 3 (1804). Again, since the contents of the diagnosis result list 1801 and the list backup 1802 do not match, the execution of the third loop is determined. Even in the third loop, the contents of the diagnosis result list are first copied to the list backup. And although all the diagnostic rules were applied with respect to the object of a diagnostic result list | wrist, the diagnostic rule which fires does not exist and a new diagnostic result was not obtained. In this case, the contents of the diagnosis result list 1801 and the list backup 1802 match. Thereby, it is considered that the final diagnosis result has been issued, and the infinite loop is terminated.

  In this example, the contents of the diagnosis result list 1801 and the list backup 1802 are confirmed to match / not match, but the number of results (judgment result and diagnosis result) listed in the both are compared and the two match. It may be configured to end the infinite loop (which can be interpreted as a case where the number of results listed in both is constant).

  In the infinite loop processing from step 1604 to step 1615 in FIG. 16, first, the diagnosis result list is copied to the list backup (step 1605). Then, a loop process for applying all the diagnostic rules to the target (determination result and diagnosis result) in the diagnosis result list is performed. The range from step 1606 to step 1612 is a corresponding loop. This loop is called loop 2.

  In the range of loop 2, the diagnostic rules not applied in loop 2 are applied to the diagnostic result list one by one, and when the diagnostic rule is fired, a new diagnostic result is acquired (step 1608). Details of processing in the range of (A) 1607 to (B) 1609 in FIG. 16, that is, details of step 1608 will be described later in detail with reference to FIG. By the way, in step 1608, even if a diagnostic rule is applied, if a fire does not occur, a new diagnostic result is not acquired. In this case, the new diagnosis result may be interpreted as an empty diagnosis result.

  In step 1610, it is determined whether or not the new diagnosis result acquired in step 1608 is included in the diagnosis result list. If not included, the new diagnosis result is added to the diagnosis result list (step 1611). Then, the presence / absence of an unapplied diagnostic rule is checked. If there is an unapplied diagnostic rule, the diagnostic rule is applied to the diagnosis result list (step 1608). If there is no unapplied diagnostic rule, the process goes to step 1613. move on. On the other hand, if it is determined in step 1610 that it is included, the presence or absence of an unapplied diagnostic rule is confirmed without adding the new diagnostic result, and the process proceeds to step 1608 or step 1613 depending on the result. .

  When the loop processing for all diagnosis rules is completed, it is determined whether the diagnosis result list matches the list backup (step 1613). If they match, the diagnosis result is output and the process ends (step 1614). If not, the process returns to step 1604 and loop 1 is repeated.

  Here, the details of the processing in the range from (A) 1607 to (B) 1609 in FIG. 16 (that is, details of step 1608) will be described using FIG. This process is a process for determining firing when the diagnostic rule being selected in loop 2 in FIG. 16 (herein referred to as “selected diagnostic rule”) is applied to the diagnostic result list. In FIG. 17, the determination result and the diagnosis result may be collectively referred to as “result”.

  First, in the range from step 1701 to step 1708, processing is performed for all conditions (the number of conditions is n) in the condition part of the selective diagnosis rule.

  First, in step 1702, the all ignition determination flag is initialized. The all ignition determination flag is information indicating whether the condition is satisfied, and the initial value of the all ignition determination flag is set to zero. Zero means not satisfied and 1 means the condition is satisfied.

  In the range from step 1703 to step 1707, processing is performed for all determination results and diagnosis results of list backup. The reason why the list backup is used here is that the diagnosis result list may increase the diagnosis results within the range of loop 2 in FIG.

  First, an unselected result is selected from the list backup, and it is determined whether or not the result matches the condition item (step 1704). If they do not match, another result not selected in the list backup is selected, and the process for the result is advanced.

  If it is determined in step 1704 that they match, it is next determined whether or not the result state satisfies the condition state range (step 1705). If not satisfied, select another unselected result in the list backup, and proceed with the result. If satisfied, 1 is set to all the ignition determination flags corresponding to the conditions in the condition part.

  When the above processing is completed for all the results of the list backup, it is determined whether the condition is satisfied for the next condition included in the condition part of the selection diagnosis rule.

  When the above process is completed for all conditions in the condition part of the selective diagnosis rule, the process proceeds to a process in the range from step 1709 to step 1712 to determine firing. First, in the range from step 1709 to step 1711, it is determined whether all ignition determination flags are 1 for all conditions (step 1710). If the total firing determination flag for all conditions is 1, a new result is generated based on the result part of the selective diagnosis rule and the state of the result, and the process proceeds to (B) 1609. The new result generated here is a diagnostic result. On the other hand, if even one of the conditions is not satisfied, that is, if a condition in which all the ignition determination flags are zero is included, the process immediately proceeds to (B) 1609.

  Next, an example of diagnostic processing is shown. The determination results of oil test items (assuming that each determination result is composed of a combination of (item, direction, state)) are (TAN, upper, Urgent), (TBN, lower, Critical), (IRON, upper, (Critical), (COPPER, upper, Alert), (ALUMINIUM, upper, Alert), (NICKEL, upper, Alert), (OIL_USE_TIME, lower, Critical) are obtained. Here, TAN (Total Acid Number) means the total acid value, and TBN (Total Base Number) means the base number. IRON is iron, COPPER is copper, ALUMINIUM is aluminum, and NICKEL is nickel. OIL_USE_TIME is the oil usage time, and the lower (lower) Critical means that the usage time after the oil change is short.

  When the flowchart of FIG. 16 is started, all the determination results are added to the diagnosis result list. An example of the diagnostic rule is shown in FIG. Each diagnosis rule shown in FIG. 19 includes, as its IF part (condition part), a “negative instruction” that is an identifier that is determined to be true when the condition is not satisfied, the name of the test item, the contents of the diagnosis result, the symbol or When the “item” including the designation and the “item” is the name of the test item, the “direction” for identifying whether the judgment direction is the upper side or the lower side, and the standard abnormality level Specify the target substance when the "status" meaning, the "status range" that specifies the range of the corresponding error level based on the abnormal level of the "status", and the "item" are the contents of the diagnosis result "Argument item" for. In addition, as the THEN part (result part), the result is denied as “diagnosis category”, which means the classification of the diagnosis object such as the property, metal wear powder, external contaminants, additives, and the judgment result of the oil test value "Negative instruction" that means an identifier, "Result" that is the content, symbol or designation of the diagnostic result, and "Target" to specify the target substance for the diagnostic result expressed in the "Result" Argument item ". Note that the diagnostic rule shown in FIG. 19 is merely an example, and it is sufficient that at least one of the items listed above is included in the IF part or the THEN part, and a configuration other than those listed above may be provided. I do not care.

  In the first iteration of loop 1 in FIG. 16, all the diagnosis rules in FIG. 19 are first applied to the above determination result. As a result, the diagnosis rule No. 1, 2, 3, 4, 5 and 6 are ignited, and as a new diagnosis result (here, composed of a combination of (diagnosis result (reference item) and state)), , Urgent), (Low base number, Critical), (Abnormal wear (IRON), Critical), (Abnormal wear (COPPER), Alert), (Abnormal wear (ALUMINIUM), Alert), (Abnormal wear (NICKEL), Alert) ) Is obtained. The obtained diagnosis result is added as a new result to the diagnosis result list.

  In the second iteration of loop 1, all diagnosis rules are applied to each result in the diagnosis result list at the end of the first time. For the diagnostic rule fired at the first time, the diagnostic result by the diagnostic rule has already been obtained, so a new diagnostic result is not generated (that is, the diagnostic rule that has already been fired is excluded from the application target after the second time). May be). In the second iteration, no. The diagnostic rules 7 and 8 are ignited, and (thermal oxidation deterioration, critical) and (“abnormal wear of piston and piston ring”, critical) are obtained as new diagnostic results. The obtained diagnosis result is added to the diagnosis result list.

  In the third repetition of loop 1, all diagnosis rules are applied to each result in the diagnosis result list at the end of the second time. This time No. Nine diagnostic rules are fired, and a new diagnostic result (combustion gas leakage from the piston, Critical) is obtained. The obtained diagnosis result is added to the diagnosis result list.

  All the diagnosis rules are similarly applied to the fourth repetition of loop 1, but no new diagnosis result is obtained. That is, since the contents of the diagnosis result list and the list backup are the same, loop 1 ends here.

  FIG. 21 is a diagram showing a chain structure obtained by linking the determination result and the diagnosis result that appear in the above-described example of the diagnosis process with the diagnosis rule of FIG. In this figure, the final diagnosis result is enclosed by a square box, and the determination result is underlined. Further, the encircled numbers in the figure correspond to the No (number) of the diagnostic rule in FIG. As is clear from this figure, a complex logical structure caused by "thermal oxidation deterioration" and "abnormal wear" is obtained for the diagnosis result "combustion gas leakage from the piston" by the above diagnosis process. You can see that it is obtained.

  Next, the chain structure (linkage relationship) obtained by the diagnostic process will be described. As described above, in the diagnosis process, the result of the condition part of the fired diagnosis rule and the diagnosis result of the result part can be associated with each other, and if this is connected, a chain structure can be formed by the determination result and the diagnosis result. FIG. 20A shows the relationship between one diagnosis rule and the diagnosis result. Since the diagnosis rule stipulates that one result is obtained for one or more conditions, when one or more diagnosis results to2002 satisfy the condition of the diagnosis rule rule 2001, the diagnosis result from2003 is obtained. Will be.

  An example of the linkage relationship obtained by the diagnostic processing is shown in FIG. The chain relationship obtained by the diagnosis process always starts from the determination result 2013 indicated by a black square, and is finally shown by a square with a cross (diagonal line) through zero or more diagnosis results 2012 indicated by a white square. It always ends with the diagnosis result 2011. Therefore, the final diagnosis result can be hierarchized by setting the hierarchy to 0 and increasing the hierarchy toward the determination result. The simplest linkage relationship is one in which one final diagnosis result is obtained from one determination result (1) simple linkage). Further, since the diagnosis rule can have two or more conditions, there are a plurality of causal diagnosis results as in 2), and there are also diagnosis results that are linked to these via the diagnosis rule. 3) As shown in the chain when there is the same cause (2014) for the two final diagnoses, one diagnostic result 2014 may be a condition for two diagnostic rules, rule 10 and rule 11. However, the result obtained from one diagnosis rule is one and does not overlap with the results of other diagnosis rules. Therefore, the same diagnosis result or determination result does not result even when the hierarchy increases from the two diagnosis results. For this reason, even when there are a plurality of final diagnosis results, the final diagnosis results are always obtained by different diagnosis rules. Since there are no two or more diagnosis rules that have the same condition and obtain different results, the final diagnosis results have different logical structures.

<5. Documenting the logical structure leading to the final diagnosis result>
A description will be given of documenting a logical structure from one or more determination results to one final diagnosis result. In the present embodiment, one diagnosis result (including a final diagnosis result) is obtained from one or more determination results via a diagnosis rule, and one or more diagnosis results (a target diagnosis result is obtained from the determination result). One diagnosis result (including the final diagnosis result) is obtained from the diagnosis rule (including not only those obtained but also those obtained from the diagnosis results). That is, the diagnosis rule has a function of connecting one or more determination results and / or a combination (condition) of one or more diagnosis results and one diagnosis result (result) corresponding thereto. Therefore, in this embodiment, one or more determination results and one or more diagnosis results (including the final diagnosis result) used from one or more determination results to one final diagnosis result are used at that time. Sentences are generated by using the structure generated as a result of connecting with diagnosis rules.

  FIG. 22 shows an example of a chain relationship (a) connecting one or more determination results and one final diagnosis result, and a tree structure (b) generated by the chain relationship. In the present embodiment, the diagnosis rule derives one diagnosis result (result) from a predetermined condition (specifically, a combination of a predetermined determination result and / or diagnosis result). In the illustrated (a) chain relationship, a circle figure indicates a diagnosis rule, and one or more determination results and / or one or more diagnosis results that are “conditions” are combined with one diagnosis result that is “results”. Yes.

  The tree structure is obtained by linking conditions and results associated with each other via a diagnosis rule in a chain relationship, and the determination results and diagnosis results that are the conditions and results are nodes. That is, (a) a diagnosis rule is removed from the chain relation and linked by a link becomes (b) a tree structure.

  The nodes constituting the tree structure have a parent-child relationship according to the top and bottom of the hierarchy (hierarchical number size). Among adjacent hierarchical nodes, a node in an upper hierarchy (a hierarchy having a relatively low hierarchy number) is a parent, and a node in a lower hierarchy (a hierarchy having a relatively high hierarchy number) is a child. For a node in a hierarchy, it is a parent if it is above that hierarchy and a child if it is below it. For example, for a node in hierarchy 2, a node in hierarchy 1 is a parent and a node in hierarchy 3 is a child.

  In the tree structure according to the present embodiment, the final diagnosis result is a node (root node) having no parent, the determination result is a node (leaf node) having no child, and the final diagnosis result (root node) and the determination result ( A diagnosis result existing between (leaf nodes) is a node (internal node) having both a child and a parent. In the following, for simplicity, the root node may be referred to as the root, the leaf node as the leaf, and the internal node as the node.

  The tree structure represents a logical structure that connects the determination result and the final diagnosis result, and causes are connected in the direction from the root to the leaf. Therefore, in the present embodiment, a sentence is generated by circulating a tree structure and sequentially connecting phrases and sentences from the leaf that is the first cause of the logical structure toward the parent. Here, a character string that ends with a punctuation mark (,) is a phrase, and a character string that ends with a punctuation mark (.) Is a sentence.

FIG. 23 shows a procedure for converting the example of FIG. The numbers in circles in the figure indicate the order in which phrases or sentences are output. First, start from the route that is the final diagnosis result, and move toward the leaf to the VISCOSITY 40 with a kinematic viscosity drop of 40 ° C. When the leaf arrives, the phrase “VISCOSITY 40 is Critical to lower” is output. Next, the process proceeds to a smaller hierarchy (parent node) and outputs the sentence “Kinematic viscosity is reduced by 40 ° C. is critical.” When proceeding to a lower hierarchy, the route returns to the root which is a node (branch point) having another child node. Therefore, it goes to another child node and goes around toward one leaf that can be traced with mixing of different oils, increasing additives, and MOLYBDENUM. When it reaches the leaf, it outputs the phrase “MOLYBDENUM is upper, Critical”. Next, the process proceeds to a smaller hierarchy, and the sentence “Additive increase is critical” is output, and the sentence “Different oil mixture is critical” is output. Returning to the previous branch point, there is no further branch. Since the branch point is also a route, the patrol is completed. Therefore, finally, “Critical is a mixture with low viscosity oil, rather than a kinematic viscosity drop of 40 ° C.
To summarize the above output, “VISCOSITY 40 is critical for lower, and kinematic viscosity is 40 ° C. is critical. MOLYBDENUM is critical for upper, and additive is critical. , And by mixing different oils, the low viscosity oil mixing is critical. "

  Sentence generation is basically a computer by using a method of generating a sentence by outputting a phrase and sentence for each node toward the root direction by outputting a phrase and sentence when the tree structure is reached and the leaf is reached. 114 can be processed. In the case of a route or node with a branch, when all the children of the branch have been visited, a sentence is output collectively. When this is generalized, phrases and sentences are output in six ways as shown in FIG.

  “Numbers in parentheses” in FIG. 24 indicate determination results or diagnosis results, “Status” indicates a state, and “Dir” indicates a direction. In addition, the connection in the case where the relationship between the parent and the child is 1: 1 is called a single chain, and the relationship of 1: n (n: 2 or more) is called a double chain.

  (A) A single-chain leaf (determination result) 2401 outputs the phrase “(test item) is (direction) in (state)” indicating the determination result. (B) At a single-chain node (continuous single-chain), a sentence “(diagnosis result) is (state)” indicating the diagnosis result is output. In the illustrated example, sentences are output in the order of nodes 2404, 2403, and 2402. (C) In a single-chain route (final diagnosis result) 2405, “So (diagnosis result) is (state)” is output as a summary sentence. (D) In the double-chain node 2406, a sentence “(child diagnosis result 1), (child diagnosis result 2),... ), (Diagnosis result) is (status). " The underlined portion in the sentence of FIG. 24D may be called a “child enumeration phrase”. (E) In the double-stranded route 2407, first, a plurality of children (nodes, leaves) are gathered, and further sentences are put together into sentences “(child diagnosis result 1), (child diagnosis result 2),... From (child diagnosis result n), (diagnosis result) is (state). " In (f) double-stranded leaf (judgment result) 2408, if a phrase is output as in (a) single-chain leaf (judgment result), it ends with a punctuation mark, so the sentence is connected to the next output. Therefore, the phrase “(test item) is (state) in (direction)” that ends with a point is output. The output at each route, node, and leaf is stored in a variable (memory area) in the computer 114, and finally output as a sentence to a file or the like collectively.

  The above is the output contents of phrases and sentences for the root, nodes, and leaves of the tree structure.

  The sentence generation process of the diagnosis result is shown using the flowchart of FIG. The overall flowchart of FIG. 25 (a) allows a plurality of routes as final diagnosis results to be present, and indicates that sentences are generated for each route for all routes (final diagnosis results). The text is generated while circulating through the nodes (root, node, leaf) of the tree structure related to the route selected by recursive call. Since phrases and sentences are connected to each of the root, node, and leaf, one recursive call is defined as a unit diagnosis sentence generation process. Loops for all routes in the range from step 2501 to step 2503 are set, and the unit diagnosis sentence generation processing 2502 is called for each route.

  FIG. 25B shows a flowchart of the unit diagnosis sentence generation process. The processing is executed by the computer 114. The process starts from step 2511. First, the number of children for the case where the root, node or leaf to be called is a parent is determined. If 0, the call target is a leaf; if 1, it is a single-chain route or node; if greater than 1, it is a double-chain route or node.

  If the number of children is 0 in step 2511, the number of children of the parent when the call target is a child is determined (step 2521). If the number of children of the parent is 1 in step 2521, since the leaf to be called is a single-chain leaf, a phrase corresponding to the single-chain leaf is generated and used as an output character string (step 2522). If it is greater than 1 in step 2521, the call target leaf is a double-stranded leaf, so a sentence corresponding to the double-stranded leaf is generated and used as an output character string (step 2523).

  If the number of children is 1 in step 2511, first, unit diagnostic text generation processing is performed for the child node or leaf (step 2531). The call recursively traverses the tree structure to the leaf. When the unit diagnostic sentence generation processing 2531 for the child is completed, the number of the hierarchy to which the calling target belongs is determined (step 2532). If the hierarchy is 0 in step 2532, the call target is the root. Therefore, a sentence corresponding to the single-chain route is added to the output character string (step 2533). On the other hand, if the hierarchy is greater than 0 in step 2532, the call target is a node. Therefore, a sentence corresponding to the single-chain node is added to the output character string (step 2534).

  If the number of children is larger than 1 in step 2511, the call target is a double-stranded parent, and unit diagnostic sentence generation processing is performed for all children through a loop from step 2541 to step 2543 (step 2542). When the unit diagnostic sentence generation process for all children is completed, a child enumeration phrase adding process for enumerating the child diagnosis results or determination results to generate a phrase is performed (step 2544). Next, the number of the hierarchy where the call target exists is determined (step 2545). If the hierarchy is 0 in step 2545, a sentence corresponding to the double-chain route is added to the output character string (step 2546). On the other hand, if the hierarchy is greater than 0, a sentence corresponding to the double-stranded node is added to the output character string (step 2547).

  Finally, the output character string is returned (returned) to the caller, and the unit diagnostic sentence generation process is terminated (step 2561). The above is the description of the sentence generation process of the diagnosis result.

  In addition, the sentence shown above only connects the diagnosis result and the state with particles etc. for each diagnosis result, but according to the combination of the diagnosis result and the state, it is only necessary to assign phrases and sentences to different expressions. Can be converted into a more natural style.

<6. Creation of report>
The creation of a report displayed on the display device 180 (monitor) and output or printed as necessary will be described. An example of the report format is shown in FIG. The report includes a header portion 2601 for displaying information for identifying customers and machines and information for identifying the report, an oil use information portion 2602 for displaying oil use information, and an oil test item determination. It is used to create a diagnostic result in a determination result part 2603 for displaying a result, a diagnostic result part 2604 for displaying a diagnostic result, a comment field 2605 for displaying a diagnostic text and a recommended text, and a comment field 2605. A tree structure display field 2606 for displaying a tree structure indicating a logical structure from the determination result to the final diagnosis result is provided.

  The determination result unit 2603 uses a format that changes the display color or mask (hatching) according to the determination result of each test item. The rank of the determination result is shown in three stages, Normal, Category, and Critical.

  In the above description of the determination process, the rank is set to four levels, Normal, Alert, Urgent, and Critical. However, it is not necessary that the rank classification of the determination process and the report is the same. A configuration may be adopted in which an item determined as “Urgent” is displayed as “Cation” on the report. If the number of ranks in the determination process is larger than the number of ranks in the report, it can be assigned.

  The description content of the text in the comment field 2605 in the report will be described with reference to FIG. In the format of FIG. 27, sentences are mainly divided into an opening part 2701, a diagnosis part 2702, and a recommendation part 2703.

  The opening part 2701 describes version information of information used for processing in the program, such as management reference values, diagnosis rules, and recommended rules stored in various databases 121 to 125, and notes related to each information. By registering version information as records in the databases 121 to 125, these pieces of information can be easily and automatically output as sentences by calculation processing. The sentence may be, for example, “Management standard is Ver. (Version number). Diagnosis rule is Ver. (Version number)”.

  Diagnosis text is written in the diagnosis unit 2702. The diagnostic text is divided into a property part 2711, a metal wear powder part 2712, an external contaminant part 2713, and an additive part 2714. When a plurality of final diagnosis results are obtained, a diagnostic sentence is created for each final diagnosis result. As an example of display in that case, there is a display in which the diagnosis text is divided into paragraphs for each final diagnosis result.

  In addition, the diagnostic text is data of diagnostic classifications in the result part of the diagnostic rule that led to the final diagnostic result (specifically, Property (property), Wear (metal wear powder), Ingress (external contamination), Additive ( According to the additive)), it is arranged in the parts 2711, 2712, 2713, 2714 corresponding to the respective diagnostic category. As the processing of the computer 114, there is a method of outputting diagnosis text corresponding to the final diagnosis result for each paragraph in the order of properties, metal wear powder, external contaminants, and additives. However, this is merely an example, and the order of processing and the order of small sentence output can be changed as appropriate, and the diagnostic texts of some diagnostic categories can be omitted.

  In the recommendation unit 2703, a sentence (recommended sentence) describing an action (recommended correspondence) that a user, a construction machine manufacturer, or an agent desirably performs in view of the diagnosis result is described.

  A recommended sentence (recommended correspondence) exists for each diagnosis result, and the diagnosis result and the recommended sentence are associated with each other through a recommendation rule that associates both. As the format of the recommended rule, the same IF-THEN rule as the diagnostic rule is used, the diagnosis result and state are set in the condition of the “condition part”, and the recommended text is set in the result of the “result part”. Note that the recommended text may be assigned to only some diagnosis results.

  FIG. 28 shows an example of recommended rules. The setting of the recommended rule is the same as the above-described diagnosis rule. In the recommended rule, since only the diagnosis result is set in the “condition part” (the determination result is not a condition), setting of the direction is unnecessary. Also, the recommended text is directly set in the “result section”. In the case of the diagnostic rule, the diagnostic result is obtained from the result part, but in the case of the recommended rule, the recommended sentence is obtained from the result part as well. The data structure of the recommended text is the same as the data structure of the diagnosis result shown in FIG.

  The process of outputting the recommended text will be described with reference to the flowchart of FIG. This processing is performed by the computer 114. First, the final diagnosis result of the model and device to be determined is acquired (step 2901), and the recommended rule for the model and device to be determined is acquired (step 2902). The recommended text is obtained according to the final diagnosis result by firing the recommended rule. Since multiple recommended sentences may be obtained, the recommended sentences are stored in the recommended sentence list. At this time, as an initialization process, the recommended sentence list is emptied (step 2903).

  In the range from step 2904 to step 2909, all recommended rules are looped. First, the recommended rule is applied to the recommended sentence list, and when fired, a new recommended sentence is acquired (step 2906). The processing in step 2906 corresponds to the processing in step 1608 in FIG. 16, and the range from (A) 2905 to (B) 2907 corresponds to each processing shown in FIG. More specifically, if “list backup” in step 1703 in FIG. 17 is read as “recommended sentence list”, each process in FIG. 17 corresponds to a range from (A) 2905 to (B) 2907.

  When step 2906 is completed, it is added to the recommended sentence list (step 2908). However, it is not necessary to add if the new recommended text is not obtained. If all the recommended texts obtained are output, the final recommended text can be obtained. The above is the description of the recommended process.

<7. How to proceed with work using this embodiment>
A method of advancing work in the present embodiment when diagnosis is automated will be described. FIG. 30 shows a sequence diagram of the exchange of the oil diagnosis service by the customer, the construction machine manufacturer or its local sales agent, and the oil analysis specialist company. Here, customer means a user of a construction machine (specifically, the owner, manager, driver, maintenance vehicle, etc. of the construction machine), and the service department of the construction machine manufacturer or its local sales agent. Service representatives and salespersons are included.

  The construction machine manufacturer owns the computer 114 according to the embodiment of FIG. The DBs 121, 122, 123, 124, and 125 are preferably owned by a construction machine manufacturer, but are not essential.

  The construction machine manufacturer concludes a contract with the oil analysis specialist company, which has decided to promptly provide the construction machine manufacturer with the result of the oil test conducted by the oil analysis specialist company (oil test result). In order to complete the oil diagnosis at an early stage, it is desirable that the oil analysis specialist company notifies the construction machine manufacturer of the test results as soon as the oil test is completed. If such an agreement is concluded, an oil analysis specialist company will have the opportunity to conduct an exclusive oil analysis of the target machine for this service.

  The oil analysis specialized company owns terminals such as the oil test apparatuses 101 and 103, the oil test apparatus 102, and the computer 112 according to the embodiment of FIG. Data on the results of the oil test conducted by the oil analysis specialist company is temporarily stored in these terminals and then output to the computer 114 of the construction machine manufacturer. The oil analysis specialized company stores the oil test result in the test result DB 121 owned by the construction machine manufacturer or the oil analysis specialized company, and the computer 114 of the construction machine manufacturer acquires the data of the oil test result from the test result DB 121. Also good. In addition, the aspect of providing the data of the oil test result from the oil analysis specialized company to the construction machine manufacturer may be any aspect as long as the data can be used by the computer 114 of the construction machine manufacturer.

  Meanwhile, a construction machine manufacturer customer enters into a contract with a construction machine manufacturer or a local agent to provide an oil diagnostic service. Oil required for providing the oil diagnosis service is collected by a construction machine manufacturer, a local sales agent of the manufacturer (sometimes referred to as an agent), or a customer, and has signed a contract. Sent to an analysis specialist company. When the oil is collected by the customer himself, an oil sampling kit may be provided to the customer, and the owner, manager, driver or mechanic of the construction machine may use the kit. In addition, when a construction machine manufacturer or a local agent carries out oil collection, it may be carried out by a service person in the service department or a salesperson. The oil collector does not need to be performed by a specific person, and even when the oil collector is not limited to a specific person, it is not necessary to limit the oil collector to one person.

  The construction machine manufacturer that has received the oil test result data from the oil analysis specialist company uses the computer 114 to automatically execute diagnostic processing.

  The result of the diagnostic process is provided in the form of a report to the customer, the construction machine manufacturer or its agent. Customers include construction machine owners, managers, drivers, maintenance personnel, etc., and construction machine manufacturers or their agents include service personnel, salespersons, and the like. Each is not limited to one person, and it is not limited to everyone. The report may be provided to an oil analysis specialist company, and may be provided to another person (for example, a customer) via the oil analysis specialist company.

  The report can also be provided as electronic data. In this case, from the computer 114 that generated the report, the construction machine owner, manager, operator and mechanic, as well as the construction machine manufacturer, the local sales agent of the manufacturer, the service personnel and the sales staff of the manufacturer May be output indirectly or directly to at least one of the terminals owned by.

  By providing the report, the customer can decide to stop the construction machine before it breaks down or breaks down. A construction machine manufacturer or an agent can prepare a replacement part for which there is a high possibility of a failure based on the diagnosis result. A service person or salesperson of a construction machine manufacturer or an agent inspects the construction machine according to the diagnosis result (report), and if it is determined that repair is necessary, the construction machine maintenance is performed. When a construction machine manufacturer or an agent exchanges parts and sells new parts for the replacement, the contract may be paid for by the customer as a price, or a guarantee of free replacement It may be a service contract. Parts removed as a result of parts replacement can be collected by a construction machine manufacturer or an agent. The collected parts may be recycled after appropriate repairs or purchased for a fee.

  The above is the method of determining the oil test result according to the present embodiment, the method of setting the management standard for the determination, the method of setting the statistical management standard value, the diagnostic method, the documentation of the diagnostic result, the automatic report generation, and It is a business model.

  In the above description, all the determination results acquired from the test results and all the diagnosis results appropriately generated from the determination results via the specific diagnosis rule are all stored in the diagnosis rule DB 124. A tree structure was obtained by repeatedly applying diagnostic rules and referring to the information of the diagnostic rules fired during that time to link the corresponding judgment results and diagnostic results, but a specific combination of diagnostic results appears in the test results If it is configured to automatically generate / output a tree structure whose root is the final diagnosis result corresponding to the combination, or a diagnostic or recommended sentence corresponding to the tree structure, It does not matter if it is configured. For example, if a diagnosis result, a tree structure, a diagnosis sentence, and the like generated from the combination of all the diagnosis results can be set in advance, a management system having such a configuration can be constructed.

  DESCRIPTION OF SYMBOLS 114 ... Computer, 121 ... Test result DB, 122 ... New oil information DB, 123 ... Management reference DB, 124 ... Diagnosis rule DB, 125 ... Recommended rule DB, 131 ... Oil diagnosis system, 204 ... Statistical setting of management reference value 211 ... Oil test result, 212 ... Oil usage information, 214 ... Oil test result determination, 215 ... Determination result (oil state), 222 ... Diagnosis, 223 ... Diagnosis result, 224 ... Text generation, 232 ... Recommended, 241 ... report

Claims (11)

A management unit for the test items of oil collected from construction machinery, a diagnostic rule, and a recording unit for recording a logical configuration for performing a final diagnosis by linking the diagnostic results;
An oil test result acquisition unit that acquires a test result of the oil collected from the construction machine,
An evaluation value is obtained from the test result, and a determination processing unit that determines the degree of abnormality based on the evaluation value and the management standard,
Based on at least one of the evaluation value and the determination result of the degree of abnormality , the diagnostic rule and the logical configuration, the construction machine or a component thereof is diagnosed, and the diagnostic rule from the evaluation value to the diagnostic result and A construction machine management system including a diagnosis processing unit that generates an explanatory text or a logical configuration diagram of the diagnosis result based on the logical configuration. In claim 1,
In the setting of the management standard,
Double logarithm conversion is performed on the test values of the test results performed in the past,
Calculate the average and standard deviation of the values after the double logarithmic transformation,
Using the probability model of occurrence of the value after the double logarithmic conversion as a normal distribution, the value after the double logarithmic conversion with respect to a preset cumulative probability value is calculated,
A double exponential conversion is performed on the value after the double logarithmic conversion to calculate a value corresponding to the cumulative probability value,
A value corresponding to the calculated cumulative probability value is set as a management criterion for a certain test item among the management criteria.
A construction machine management system characterized by that. In claim 1,
In setting the management criteria of a certain test item in the management criteria ,
Double logarithmic conversion is performed on the quantitative values of the test results conducted in the past,
In order to make the probability model of the quantitative value generation after the double logarithmic transformation a Weibull distribution, that is, a double logarithmic Weibull distribution,
Arranging the quantitative values after the double logarithmic conversion in ascending order to determine the rank, and dividing the rank by the total number of data, calculate the cumulative probability value of the occurrence of the quantitative value after the double logarithmic conversion,
Using the value obtained by logarithmically transforming the quantitative value after the double logarithmic transformation as an explanatory variable and the value obtained by transforming the cumulative probability of occurrence based on the Weibull cumulative distribution equation as the objective variable, the coefficient and intercept of the linear equation are set to a minimum of 2 Calculated by multiplication,
Using the coefficient of the linear expression as the shape parameter of the Weibull distribution, the scale parameter of the Weibull distribution is calculated from the coefficient of the linear expression and the intercept,
From the double logarithmic Weibull distribution, a quantitative value after double logarithmic conversion for a preset cumulative probability value is calculated,
Double index conversion is performed on the calculated quantitative value after the double logarithmic conversion to calculate a quantitative value corresponding to the cumulative probability value,
A quantitative value corresponding to the calculated cumulative probability value is set as a management standard for the certain test item;
A construction machine management system characterized by that. In claim 1,
A construction machine management system, wherein the test items include a test of metal wear powder and external contaminants contained in oil. In claim 1,
The test includes four types of test items: oil properties, metal wear powder mixed in oil, external contaminants mixed in oil, and oil additives.
In the determination process in the determination processing unit ,
Among the test items categorized as the above properties, for the test items whose evaluation value may increase or decrease with respect to a certain value, the degree of abnormality is determined on both sides (both upper and lower sides) of the certain value. For a test item whose quantitative value increases with respect to a certain value, the degree of abnormality is determined above the certain value, and for a test item whose evaluation value decreases with respect to a certain value, Determine the degree of abnormality at the bottom,
The test item classified as the metal wear powder, the degree of abnormality is determined above a certain value,
The test items classified as external contaminants determine the degree of abnormality above a certain value,
Among the test items categorized as the oil additive, for test items that have been confirmed to be added with new oil, the degree of abnormality is determined on both sides of a certain value, and for test items that have not been confirmed to be added with new oil Determines the degree of abnormality above a certain value,
A construction machine management system characterized by that. In claim 1,
In the determination process in the determination processing unit ,
The evaluation value, the test value itself of test items or construction machine management system, characterized in that the difference between the test value and the new oil value of the test item. In claim 1,
As a management standard value of the management standard ,
Set the magnification for the new oil value ,
Or, after setting the reference value, set the magnification for the reference value.
Alternatively, the numerical value corresponding to the difference between the test value of the test item and the new oil value is directly set.
Alternatively, after setting the upper limit value or lower limit value of the management reference value, the magnification of the difference between the upper limit value or lower limit value and the new oil value is set.
Or, setting the magnification of the difference between the upper limit value or the lower limit value and the reference value in terms of setting the upper limit value or lower limit value and the reference value of the control reference value,
A construction machine management system characterized by that. In claim 1,
In setting the management standard value of the management standard ,
The operation time of the construction machine from oil change to oil collection is the oil use time,
Set the initial value of the management standard value at the time of oil change and the amount of change in the management standard value per unit time,
A construction machine management system characterized in that a management reference value at the time of oil collection is calculated from a change amount of the initial value and the management reference value. In claim 1,
The oil test result acquisition unit acquires the test result from a terminal of an oil analyzer. In claim 1,
The logical configuration is one or more tree structures having evaluation values and diagnosis results as nodes, and each of the one or more tree structures has at least one diagnosis result of the test item as a leaf node. And having a final diagnosis result as a root node, and having another diagnosis result as an internal node existing between the leaf node and the root node,
In the process of generating the explanatory text, the diagnosis processing unit circulates a node from the leaf node toward the root node in each of the one or more tree structures, and the leaf node, the root node, Connecting the phrases or sentences previously associated with each of the internal nodes to generate the explanatory text;
A construction machine management system characterized by that. In claim 1,
The explanatory text or logical configuration diagram includes the owner, manager, operator and mechanic of the construction machine, the manufacturer of the construction machine, the local sales agent of the manufacturer, the service personnel and the sales staff of the manufacturer. A construction machine management system, wherein the system is output to at least one of the terminals.

Images