JP2907873B2 - Vehicle failure diagnosis device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a failure diagnosis apparatus for a vehicle in which a failure cause, that is, a failure site can be known based on a failure symptom of the vehicle.

(Prior Art) Modern vehicles tend to be more and more complex in structure. For this reason, when the vehicle is malfunctioning, that is, when a trouble symptom occurs in the vehicle, even a considerably experienced mechanic tends to have difficulty in identifying a part that causes the trouble symptom. . This imposes a heavy burden on quick repairs.

For this reason, recently, as disclosed in Japanese Patent Application Laid-Open No. 62-6856, an apparatus called an expert system has been proposed which can identify a part causing a failure symptom from a failure symptom. . This is because, based on past data, the causal relationship between the rise in failure and the component failure that causes the failure is stored as a failure tree, and failure symptoms are sequentially input, so that the finally failed component And displays it. By storing the probability of the causal relationship between the failure symptom and the component failure that causes the failure symptom, it is possible to more accurately find the failed component.

(Problems to be Solved by the Invention) However, in the above-described conventional failure diagnosis apparatus for a vehicle, not only a large amount of time is required until a finally failed component is found, but also a failure symptom that is input is not sufficient. If the quality and quantity are not sufficient, an extremely large number of parts having the same failure probability will be displayed. This greatly reduces the utility value of such a device.

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a failure diagnosis device for a vehicle, which can more appropriately provide a person who maintains a part causing a failure in accordance with an input failure symptom.

(Means and Actions for Solving the Problems) In order to achieve the above object, the present invention takes the following viewpoints into consideration.

First, as data for failure symptoms, in addition to data common to each vehicle type in consideration of versatility, data for a specific vehicle type, for example, a limited-selling one or a new one that has just been released as a new vehicle , That is, specific to each model. In other words, conventional failure diagnosis data must be statistically inevitably generalized data, which is excellent in terms of versatility.However, failure diagnosis for specific models is not appropriate. Will not be. More specifically, if a certain failure symptom occurs, the faulty component that causes the failure symptom cannot be diagnosed based on the data common to each vehicle type, or although the reliability of the failure is low, for example, Even if it is the A part, if it is based on the data for the specific vehicle type, a completely different B part can be diagnosed as having extremely high failure certainty.

Secondly, according to the present invention, in addition to data indicating a causal relationship between a failure symptom and a failed component that causes the failure symptom, a functional chain system is prepared as data, and this is reported to a maintenance person, so that the failed component is notified. Even if this cannot be clearly diagnosed, the maintenance person can easily understand the parts to be inspected. This functional chain system shows a connection relation of the component assemblies with respect to a subsystem level system such as an ignition system and a fuel system, that is, an assembly of a plurality of components related to realizing a certain function of the vehicle. FIG. 15 is a block diagram showing an ignition system as an example. By looking at such a functional chain system, the maintenance person can more appropriately know the parts to be checked. The selection of this functional chain can be made in various ways. For example, a mechanic can make a manual selection, or can automatically select and report a high-degree one based on a causal relationship with a failure symptom. Preferably, if the degree (probability) at which a component constituting the function chain fails is prepared in the data of the function chain, the number of parts to be checked for one function chain is small. Will be.

The most preferred embodiment of the present invention
This is shown in FIG. That is, in FIG. 16, the causal relationship between the failure symptom for each vehicle model and the component that causes the failure, the causal relationship between the failure symptom for the specific vehicle model and the component that causes the failure, the failure symptom and the cause thereof, This is a system that uses a causal relationship with a specific assembly of components, that is, a functional chain system, and data of the functional chain system itself.

(Effect of the Invention) As described above, according to the present invention, it is possible to more accurately know the inspection site corresponding to the failure symptom.

In particular, by adopting the configuration as described in claim 1, even when the failure site to be inspected is not clear at first sight, the site to be inspected can be improved by utilizing the function chain penalty. It is possible for the maintenance person to know exactly. In addition, the use of the functional chain system can be made more convenient.

With the configuration as described in claim 2, it is possible to more accurately know the site to be inspected regardless of the vehicle type. Further, even when the failure site to be checked is not clear at first sight, the maintenance site can be more accurately known to the maintenance person by using the functional chain system. Furthermore, the use of the chain of functions can be made more convenient.

With the configuration as in claim 3, it is possible to reduce the number of parts to be inspected by utilizing the function chain.

(Example) Hereinafter, an example of the present invention will be described with reference to the attached drawings.

FIG. 1 is a block diagram showing an overall system of the present invention, and this system is configured using a computer.

In FIG. 1, the fault symptom force means A is indicated by four blocks 1, 2, 3, 4. That is, the failure symptom input means A includes a failure symptom input unit 1, a fifth storage unit 2, a question display unit 3, and an input information storage unit 4. The failure symptom input unit 1 is for a user of the present system to input a primary failure symptom. The fifth storage unit 2 stores the components causing the failure symptom and the probability of the causal relationship thereof,
It is stored in the form of a fault tree as described in the above publication. The question display unit 3 displays a question limited to the range of the fault tree stored in the fifth storage unit to a person who operates the question tree. The input information storage section 4 stores the input information, and the stored information is used as input information of a failure symptom for a portion constituting the present invention described later. That is, when the present invention is used, the above-described elements 1, 2, 3, and 4 are configured in order to limit the failure symptoms to some extent.

In FIG. 1, 5 is the first storage unit, 6 is the second storage unit, and 7 is the third storage unit.
Storage unit, 8 is a fourth storage unit, 9 is an inference unit, 10 is a selection unit, 11
Is a notification unit.

The first storage unit 5 stores, for each component, a causal relationship between the component and a failure symptom caused by the failure thereof together with the probability thereof as first failure information. It has been created based on. The second storage unit 6 also stores a causal relationship between the component and a failure symptom caused by the failure of the component together with the probability as second failure information for each specific component. It is created based on data specific to each model. The third storage unit 7 stores a large number of functional chains. This functional chain system indicates, for each of a large number of functions of the vehicle, such as an ignition system and a fuel system, a connection relationship (input / output relationship) of a plurality of components related to realizing this function. FIG. 15 is a block diagram showing an ignition system as an example. In the third storage unit 7, the degree of occurrence of a failure is stored together with the probability of occurrence for each component constituting one functional chain. 4th
The storage unit 8 stores, as third failure information, the component aggregates constituting each of the above-described functional chains and the causal relationship when the component aggregates fails, together with the probability thereof. Each of the storages 5, 6, and 8 is stored in the form of an inference tree described later.

The inference unit 9 stores the first, second, and fourth storage units 5, 6, and 8 according to the information stored in the input information storage unit 4.
To infer the parts that would have failed and their probabilities, as well as infer the assembly of parts that would have failed, along with their probabilities. Then, the inference result is notified by the notifying unit 8.

The selecting unit 10 selects, from the third storage unit 7, a function chain corresponding to the component assembly that is likely to have a fault, which is inferred by the inference unit 9, and selects the selected function chain. The notification unit 11 is notified together with the component failure probability. However, in the embodiment, the notification of the function chain system is performed only when the maintenance person cannot fully understand the search results of the first, second, and fourth storage units 5, 6, and 8. For this reason, a command signal for commanding the selection means 10 to select a function chain is output from the input unit 1.

The above-mentioned failure symptom input unit 1 is constituted by, for example, a keyboard. An external storage device, such as a floppy disk or a hard disk, is used for each of the storage units 2, 5, 6, 7, and 8 from the viewpoint that a large storage capacity is required. The input information storage unit 4 also includes the storage units 2, 5, 6,
Although it can be the same as 7 and 8, an internal storage device having a small storage capacity may be used. Question display section 3 and notification section
For 11, a CRT is used, for example, and a printer can also be used. Of course, the inference unit 9 is constituted by a CPU.

Failure symptom input means A The storage contents of the fifth storage unit 2 are schematically represented in the form of a failure tree as shown in FIG. This is a causal relationship, for example, in which no cranking occurs as a failure symptom, and the cause becomes more concrete as it goes downward, so that it eventually reaches a part that seems to be defective. Has become. Then, the degree of the causal relationship of the tree is shown by the probability. FIG. 3 schematically shows the structure shown in FIG. Such a fault tree is specifically stored as a number of rules as shown in FIG. That is, X is a failure symptom (the degree of the cause goes down the tree, there is a degree of cause), Y is the cause one level below X (the part goes down the tree, it becomes a part), and N is the probability (causality). Relationship confidence)
It is. As a certainty factor of the failure, as shown in FIG. 5, for example, roughly three levels of classification such as LOW, MIDDLE and HIGH may be used together, or only this rough classification may be used.

FIG. 12 is a flowchart showing the procedure until information is stored in the failure information storage unit 4 in FIG. 1, and FIG. 12 will be described below. First, in P (step) 1, a leading rule (a rule in which A in FIG. 3 is the highest order) is fetched. Next, at P2, X in the rule is looked up to search for any other rule having this X. Then, among the rules having the X, the rule having the highest certainty factor N is selected.

After P2, in P3, it is determined whether or not Y in the selected rule matches the failure symptom. If they match, the current relationship between X, Y, and N is stored in the input storage unit 4 at P4. Then, in P5, Y is changed to X
(Y is updated as X of the rule), and it is determined in P6 whether there is a rule caused by X. If the determination in P6 is NO, the process ends. Also, the judgment of P6
If YES, the processes after P2 are repeated.

If the determination in P3 is NO, in P7, a rule with a lower confidence level N is selected, and then in P8, it is determined whether a similar rule still exists. If there is another rule, the process returns to P3. If there is no other rule, at P9, a rule at a lower level than the current one (the rule at the bottom of the fault tree) is stored in the input storage unit 4. It should be noted that the time when this P9 is reached is when the part that eventually causes the failure symptom cannot reach a sufficiently lower level of the failure tree.

When information is stored in the storage unit 4 using the above-described fault tree, if a questionnaire sheet in which question items are summarized to some extent along the fault tree is created and input, the trouble can be saved. Become. However, in this case, the questions should be limited to ranges that can be easily determined, such as visual inspection and basic inspection.

First, Second, and Fourth Storage Units 5, 6, and 8 First, the vehicle configuration is classified from a large concept into a small concept as shown in FIG. That is, vehicles are roughly classified into large concepts (system level) such as engines and missions. The system level is classified into a medium concept (sub system level) such as an ignition system and a fuel system for an engine. Then, the components at each subsystem level are classified into individual component levels such as fuel pumps and injectors, for example, in regard to the fuel system.

From the above-described relationship, two types of inference trees are set: a system corresponding to a subsystem such as an ignition system and a fuel system, that is, a system for a specific assembly of components and a component level.

FIG. 7 shows an example of an inference tree for the component polymer. In FIG. 7, an ignition system is shown as an example. That is, when the ignition system fails, first, a failure symptom of a large concept of poor starting is associated with the ignition symptom. The next lower concept (failure symptom or cause) causes the ignition plug to become wet. And that the igniter is generating a false signal. In this way, the inference tree is related so that the failure symptoms that occur due to the failure of the system as a component assembly become more specific from the larger concept to the more specific concept from top to bottom. I have. Of course, the upper and lower causal relationships in the inference tree are weighted as failure probabilities.

FIG. 8 schematically shows the above-described inference tree. Specifically, such an inference tree is stored as a rule as shown in FIG. 9, similarly to the case of a fault tree. Have been. In other words, focusing on Rule 1, the lower levels B ₁ and B ₂ of A _{1 in} FIG. 8 are stored together with their conviction levels and convictions.

FIG. 10 is an inference tree for each part, and the leading part is the ninth tree in that individual parts come in place of the parts aggregate.
It is different from that in the figure (FIG. 7). Then, such an inference tree for each component is also stored as a rule together with its confidence level and confidence as shown in FIG. In addition,
In the case of an inference tree in parts, it is often the case that those at the same level above and below the inference tree correlate together to indicate one failure symptom. In this case, a separate rule is formed as AND information (FIG. 11). See Rule 3). Of course, the inference tree for each part is created as an independent tree for each vehicle type (first storage unit 5) and a specific vehicle type (second storage unit 6).

Third storage unit 7 This stores a chain of functions as shown in a block diagram in FIG. 15, and is specifically coded in the following format.

(Distributor (Function “Distribution of ignition energy”) (Connecting parts (before IG coil) (after HT code)) That is, for example, focusing on one part of the ignition system, “Distribution of ignition energy” The function performed by the component, the connection name, ie, the component name “IG coil and HT code” before and after the input / output relationship, and the failure probability of the component (distributor) are stored as one set. Regarding the IG coil, the components before and after the connection are stored as “battery and ECU” on the input side and “distributor” on the output side. Note that the failure rate is set as a relative value with respect to other components within the range of one functional chain. When the failure rate is related to a plurality of functional chains, one component is assigned to each functional chain. It is good to memorize a failure probability.

The inference unit 9, the selection unit 10, and the notification unit 11 The inference unit 9 performs the above-described inference tree for the assembly of parts and the inference tree for each component (for each vehicle model and for the specific model) in accordance with the input failure symptoms. In this case, a component assembly and a component which are likely to have a failure are inferred together with the failure probability. In other words, if there is a rule in which the content corresponding to the input failure symptom is stored, the component assembly or component fails at the confidence level and confidence probability stored in the existing rule. Infer that The inferred result is finally appropriately selected according to the correlation between the component assembly and the component, the confidence level, and the confidence, to obtain a final inference result. Of course, this final inference result
11 will be notified.

An example of the result finally inferred by the inference unit 9 is as follows.

Parts assembly Control system (HIGH 0.6) Fuel system (HIGH 0.3) Parts Water temperature sensor (HIGH 0.5) Control unit (HIGH 0.3) Injector (MIDDLE 0.8) EGR valve (LOW 0.7) In view of these results, the maintenance person can determine the correlation between the water temperature sensor, the control unit, and the injector based on the correlation between the component assembly and the component, and the confidence level and the confidence factor.
It is considered that the EGR valve is excluded from the suspected malfunction while the operator suspects that the malfunction has occurred.

Even if the failure information stored in the first, second, and fourth storage units 5, 6, and 8 is searched, the maintenance person may not be fully satisfied. At this time, by the selection means 10,
From the functional chain stored in the third storage means,
A functional chain system corresponding to the component aggregate having the highest failure probability is selected from the component aggregate estimated to have failed. Then, the selected chain of functions is notified to the notifying unit 11, and a part having a high failure probability in the chain of functions is also notified. A mechanic who sees such a chain of functions checks the output state of the chain of functions to determine whether there is an abnormality in the chain of functions. If the result of the output check is abnormal, the check is performed in the descending order of the probability of failure among the components constituting the functional chain. If the output check is normal, the function chain corresponding to the component assembly that is inferred to have the next highest probability of failure is selected, and the above-described procedure is repeated.

FIG. 13 and FIG. 14 are flowcharts showing a procedure for obtaining an inference result using the above-described inference tree. Note that, as described above, the input information storage unit 4 stores the failure symptoms input up to now and the parts considered to be failures in accordance with the failure symptoms to some extent in a limited range. Only needs to be performed based on the information stored in the storage unit 4.

First, in Q1 of FIG. 13, the processing of FIG. 12 is performed.

Next, in Q2 and Q3, it is inferred according to the failure symptom whether the component assembly has failed together with the failure probability (search in the fourth storage unit 8). In Q4 and Q5, it is inferred as to whether or not a component unit common to each vehicle type has a failure together with the failure probability (first storage unit 5).
Search). Further, in Q6 and Q7, it is inferred as to whether or not there is a failure in the component unit of the specific vehicle type together with the failure probability (search in the second storage means 6).

Note that the inference of Q6 may be performed only when there is an inference command for a specific vehicle type from the input unit 1.

In Q8, based on the inference results of the parts assembly and the parts (there are two types, common to each vehicle type and specific vehicle type), based on the comprehensive inference, ie, the correlation between each other, the confidence level, and the probability of failure, Are removed, and the final inference result after obtaining this processing is reported to the reporting unit 11 in Q9.

Details of the above-described Q2 (Q4, Q6) are as shown in FIG. First, in R1, a component aggregate or an inference tree corresponding to the component is selected (selection of the first storage unit 5, the second storage unit 6, or the fourth storage unit 8). Next, in R2, the rule located at the uppermost position is selected.

In R3, it is determined whether or not there is any information stored in the selected rule that matches the input failure symptom. If the determination in R3 is YES, in R4, the part assembly (part) that is the current inference target is stored together with the confidence level and the confidence in the rule determined as YES in R3. Thereafter, in R5, the rule is updated to the next lower level rule in the inference tree, and the processing in and after R3 is repeated.

If the determination in R3 is NO, in R6 another rule with the same level in the inference tree is selected, and then it is determined in R7 whether such a rule exists. When the determination in R7 is YES, the process returns to R3, and when the determination is NO, the process is terminated (search termination).

After Q9 in FIG. 13, it is determined in Q10 whether or not the inference result of Q9 is sufficiently satisfactory. If the determination is YES, the process ends. NO in Q10
In the case of, in Q11, the functional chain corresponding to the component aggregate that is estimated to have failed in Q3 and has the highest probability of failure among the component aggregates is selected, and the selected functional chain is selected. The notification unit 11 is notified of the failure probability of each component constituting the functional chain system. In Q12,
The maintenance person checks whether the output state of the selected function chain is normal. If there is no abnormality as a result of this check, the process returns to Q11 again, and the function chain corresponding to the component assembly having the next highest failure probability is selected and displayed.

When the determination in Q13 is YES, the inspection is performed in the order of components having the highest failure probability among the components constituting the functional chain.

Modification 1 In this type of failure symptom system, how accurately a user recognizes a failure symptom is important in inferring a failure site. From this point of view, using the question display unit 3 in FIG. 1, the failure symptoms input to the inference unit 9 (stored in the storage unit 4) as described below are more specific and accurate. It is good to put together in advance so that it becomes something.

First, as described below, a question story (primary question group) and a secondary question group are prepared in advance in accordance with the failure phenomena which are roughly classified as follows.

Failure phenomena -1. Start-up failure --- 1-1 Question story | 1-2 Secondary question group -2. Idle malfunction -2-1 Question story | 2-2 Secondary question group -3. -1 Question Story | 3-2 Secondary Question Group -4. Entrance --- 4-1 Question Story 4-2 Secondary Question Group Prepares knowledge (= next action knowledge) for deciding which next question is to be made according to the input value of the user corresponding thereto.

When a largely categorized symptom such as “defective starting” or “defective driving” is input, it is determined which question story to use. When the question story is determined, the doctor is asked according to this. At this time, since the next action knowledge is determined in advance as described above, the processing according to this next action knowledge is performed. Now, as a premise of a major failure phenomenon of "start failure", the following is described.

Form of inquiry processing Question story (code 1-2 ...---n where question is set) Secondary question group (code 1-2 ...---------------------------------- n) Code 1 (Question item Possible answer code Action code) Looking at a specific example, the question story “Incorrect starting” (Q1 Q2 Q3 Q4 Q7 Q19 Q20 Q21) Secondary question group (Q5 Q6 Q15 Q16 Q17) Question code (Q1 “Start condition is“ QU1 ACT1 ”) (Q2“ Reproduce ”QU2 ACT2) − − − − − − − − − Possible answer code (QU1 (QQ1 QQ2 QQ3)) (QU2 (QQ21 QQ22 UNDEF) ) Actual value of answer code (QQ1 “Do not crank”) (QQ2 “Crank but do not detonate”) (QQ3 “First detonate but not complete detonation”) (QQ21 “Always happen”) (QQ22 “Occasionally Only happens ”) (UNDEF“ I don't know ”) Action code (ACT1 (if QQ1 then Q5 (Else CONTINUE)) (ACT2 (if QQ21 then Q4) (if QQ22 then Q6) (else CONTINUE)) In the above example, the input result of the interview Q1 (= start state), that is, the user's answer result is QQ1 (= In the case of (No ranking), the action is Q5, so the question story is removed and the process moves to the secondary question group. User's answer is QQ
In the case of 2 (= cranking but not first explosion) and QQ3 (= first explosion but not complete explosion), since the action is CONTINUE, the interview will be continued along the question story. Similarly, in the case of the interview Q2, if the input value of the user is QQ21, the process proceeds to Q4, if the input value of the user is QQ2, the process proceeds to Q6, and if the input value is other than that, the interview along the question story is continued.

As described above, even if the user only knows a very rough failure phenomenon, a more specific and accurate failure symptom can be input when inferring the failure part using the questionnaire, and the failure diagnosis is performed. It is possible to perform more accurately.

Modification 2 A failure symptom based on the failure information common to each vehicle type described above,
Only one of the failure symptoms based on the failure information of the specific vehicle type can be executed in response to a request from the user.
In this case, as described below, the storage of the failure information and the failure diagnosis may be performed using the rules of “if = if, if” and “then = is”.

In this case, the “if” part often has a plurality of conditions (= failure symptoms), and if all conditions are not satisfied, the “th”
The rule is divided into an “AND” rule in which “ne” is not executed, and an “OR” rule in which “then” is executed if some conditions are satisfied. Such a rule is basically configured so that evaluation is performed in descending order of priority. In the “if” part (= a plurality of conditions) of each rule, “next action knowledge” that determines the next evaluation action based on the comparison result of the priority and the input value of the condition is defined.

An example of a description format of rules as described above and an example of evaluation will be described below.

Rule description format Rule 1 (RULE1 (AND) (if (P1ΔΔΔ (then eq action) (else action))) (P2 □□□ (then eq action)) (then ne action)) (then (failed part name Possible Failure form Confidence)) (AND) ---- Indicates the evaluation method.There is another OR. In the case of AND,
It is useless if all conditions do not exist, and in the case of OR, basically it is sufficient if there is something. However, when an action is set, the instruction (description) is followed.

 P1, P2-----Priority is given. The smaller the number, the higher the priority.

 ΔΔΔ, □□□ ---- Condition ~ Value to be compared with the input result.

 eq, ne --- Abbreviation of eq = equal, abbreviation of ne = not equal.

Action --------- Indicates the following action. For the value, NEXT-Compare the input information with the next condition. JUMP-Stop evaluation of this rule → go to evaluation of next rule END--End evaluation process. (Processing only when the evaluation result is very high confidence = 0.9, for example).

For example, (RULE # 10 (AND) (if (P1 QQ1 (then eq NEXT) (else JUMP))) (P2 QQ21 (then eq END) (then ne NEXT)) (then (injector airtight defect 0.95 defect))) When there are such rules, if the input values QQ1 and QQ21,
"Injector-poor airtightness-0.95" in the "then" part will be added to the list of evaluation results so far as the estimation result. On the other hand, when there is no QQ1 as the input value, the evaluation of the rule is interrupted as instructed by "else JUMP", and the evaluation moves to the evaluation of the next rule "# 11". When the rule evaluation is completed to the end, the rules are rearranged in descending order of confidence.

After the evaluation of the general rules (common rules for each vehicle) is completed in this way, if necessary, as a special case, it is determined whether or not the specific model, that is, the vehicle type rule or the new vehicle rule is not included. You only need to make an evaluation. The format of such a specific rule is as follows, where the vehicle type name and the failure symptom are “if” parts, and the “failed” part defines an estimated failure site, an estimated failure form, and a failure certainty factor.

Rule form in case of unusual case (= vehicle type rule, new vehicle rule)

[Brief description of the drawings]

FIG. 1 is an overall system diagram showing an embodiment of the present invention. FIG. 2 is a diagram showing an example of a fault tree. FIG. 3 is a diagram schematically showing FIG. 2; FIG. 4 is a diagram showing an example in which the contents of FIGS. 2 and 3 are stored. FIG. 5 is a diagram showing a setting example of a certainty level and a certainty factor. FIG. 6 is a diagram showing how the configuration of the vehicle is classified from a higher-level concept to a lower-level concept. FIG. 7 is a diagram showing an example of an inference tree for a component assembly. FIG. 8 is a diagram schematically showing FIG. 7; FIG. 9 is a diagram showing an example in which the contents of FIGS. 7 and 8 are stored. FIG. 10 is a diagram schematically illustrating an example of an inference tree for parts. FIG. 11 is a diagram showing an example in which the contents of FIG. 10 are stored. 12 to 14 are flowcharts showing control examples of the present invention. FIG. 15 is a block diagram showing an example of a functional chain system. FIG. 16 is a block diagram showing a preferred configuration of the present invention.

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-62-25234 (JP, A) JP-A-62-6858 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) G01M 17/00

Claims (3)

(57) [Claims] A first storage means for storing a causal relationship between a component and a failure symptom caused by a failure of the component together with a probability thereof as first failure information; and a function for each of a large number of functions of the vehicle. A second storage unit that stores, as a functional chain system, a connection relationship between a plurality of components related to each other, and a specific component aggregate that constitutes the functional chain system and a cause of failure of the component aggregate. A third storage unit that stores a causal relationship with a failure symptom that occurs as second failure information together with the probability thereof, an input unit that inputs a vehicle failure symptom, and a failure symptom input by the input unit. Inference means for retrieving the first failure information to infer a failed component and its probability, and retrieving the second failure information to infer a specific failed component assembly and its probability Selecting means for selecting, from the second storage means, a functional chain system corresponding to the component assembly inferred by the inference means; and selecting the inference result inferred by the inference means and the inference result by the selection means. And a notifying means for notifying the functional chain system. 2. A first storage means for storing, as first failure information, a general causal relationship common to each vehicle type together with its probability, among causal relationships between components and a failure symptom caused by a failure of the component. Of the causal relationship with the failure symptom caused by the failure of the part, the second storage means for storing the specific causal relationship unique to each vehicle type together with the probability thereof as second failure information, and a plurality of functions of the vehicle. A third storage unit that stores, as a function chain, a connection relationship between a plurality of components that are related to each other to realize the function; a specific component aggregate that constitutes the function chain, Fourth storage means for storing a causal relationship with a failure symptom caused by a failure together with its probability as third failure information; input means for inputting a vehicle failure symptom; input by the input means Based on the detected failure symptom, the first failure information is searched for a component that has a failure and the probability thereof, the second failure information is searched for a component that has a failure and the probability thereof, and the third failure information is retrieved. And inference means for inferring a particular component assembly that has failed and its probability, and selecting a functional chain system corresponding to the component assembly inferred by the inference device from the third storage means. And a notifying means for notifying the inference result inferred by the inference means and the function chain selected by the selecting means. 3. The storage system according to claim 1, wherein said storage means storing said functional chain includes a probability of failure for each specific component constituting said functional chain. The failure diagnosing device for a vehicle, wherein the notifying unit is configured to report the failure probability of each component constituting the functional chain together with the functional chain.