JPH1091473A - How to determine maintenance content for equipment - Google Patents

Abstract

(57) [Summary] [Problem] With regard to equipment such as buildings, it is possible to objectively evaluate the content of maintenance and inspection by estimating the failure rate of equipment with not only the past performance but also the actually small number of inspections,
Make it possible to set an appropriate maintenance contract. SOLUTION: In order to objectively and appropriately determine the maintenance contents of the equipment, the inspection of the equipment to be subjected to the maintenance and inspection is periodically performed regardless of the presence or absence of a failure. Divided into post-checks to be performed, a defect rate distribution by the pre-checks and a defect rate distribution by the post-checks are obtained from the results of both the pre-checks and the post-checks. Then, by multiplying the two defective rate distributions by the cost required for replacement of defective products, the respective costs of the pre-check and the post-check are calculated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for determining a failure rate distribution of equipment such as an automatic air-conditioning control device of a building and determining maintenance contents of the equipment based on the distribution.

[0002]

2. Description of the Related Art In many cases, a building owner and a building equipment seller or an inspection contractor (hereinafter referred to as an inspector) make a maintenance inspection contract and regularly inspect equipment. In addition, when a failure occurs, replacement or repair of the failed device is performed promptly. In such contracts, maintenance costs are one of the important items,
Generally, the cost is calculated based on the inspection work time and defect rate of the target equipment.

[0003] It is important to estimate the defect rate not only for equipment in a building with a maintenance contract but also for equipment in a building without a maintenance contract, that is, to estimate the defect rate. It is. Because, by knowing how much maintenance equipment inspections can reduce the occurrence of equipment failures, we can explain the necessity of maintenance inspections and promote the value judgment of maintenance inspections even for buildings that do not have a maintenance inspection contract. Because you can.

[0004] The defect rate of facility equipment in a building that has not been contracted for maintenance and inspection as described above can be known without much effort if sampling inspection is successfully performed. Also, in a building with a maintenance contract, if the failure rate of the equipment targeted for the sampling inspection is low, it is possible to reduce the cost of the maintenance inspection by omitting the inspection of other equipment.

[0005] From the above point of view, at present, the inspector who has contracted with the building side determines the defect rate of the equipment concerned from the actual value of the past inspection result of the equipment concerned and determines the defect rate on the building side. In consideration of the occurrence of troubles such as breakdowns, what kind of maintenance contract is concluded is determined by discussions between the building and the inspector.

[0006]

However, the failure rate of equipment such as a building changes year by year, and it is not appropriate to determine the failure rate only from past results. In addition, even if the inspector intends to enter into a maintenance contract, there is a problem that it is difficult to objectively show that building a maintenance contract is also beneficial to the building side.

Accordingly, an object of the present invention is to objectively evaluate the contents of maintenance and inspection of equipment such as buildings by estimating the failure rate of the equipment not only with the past results but also with the actually small number of inspections. It is an object of the present invention to provide a method for deciding the maintenance contents of facility equipment that can set an appropriate maintenance contract.

Another object of the present invention is to calculate a defect rate distribution of equipment and to calculate the inspection cost of the equipment based on the distribution of the defect rate of the equipment in order to objectively evaluate the validity of the inspection cost of the equipment. Is to provide a medium recording a program for causing a computer to execute the program.

[0009]

According to the method of the present invention, a pre-inspection for periodically inspecting equipment to be subjected to maintenance and inspection regardless of the presence or absence of a failure, and a post-inspection to be performed when a failure occurs. From the results of both the pre-inspection and the post-inspection, the failure rate distribution by the pre-inspection of the equipment and the failure rate distribution by the post-inspection are obtained, and the maintenance of the equipment is performed based on these two failure rate distributions. The content is determined.

[0010] In a specific embodiment of the present invention, the costs of the pre-inspection and the post-inspection can be calculated by multiplying each of the two defect rate distributions by the cost required to replace the defective product.

Further, the degree of failure of the equipment is divided into a complete failure for replacing the entire equipment and a partial failure for replacing parts, and in the case of a preliminary inspection, the sum of the failure rate distributions of the complete failure and the partial failure is calculated. In the case of the post-check and post-check, by obtaining the defect rate distribution of only complete defects, it is possible to obtain an actual defect rate distribution.

Further, the present invention provides a medium on which a program for recording a program for causing a computer to execute a process of determining a failure rate distribution of equipment to be subjected to maintenance inspection and determining maintenance contents of the equipment based on the distribution is obtained. There is a process to determine whether the inspection of the equipment is either a pre-inspection performed regularly regardless of the presence or absence of a failure or a post-inspection performed when a failure occurs, and the results of both the preliminary inspection and the post-inspection From the pre-inspection and the post-inspection defect rate distribution of the equipment, and calculate the pre-inspection and post-inspection costs from the two defect rate distributions and the cost required for replacement of defective products. And recording a program for calculating an inspection cost of facility equipment.

[0013]

According to the present invention, the target equipment is inspected in two modes, a pre-inspection and a post-inspection, and from each result, the failure rate by the preliminary inspection and the failure rate by the post-inspection are distributed respectively. It is estimated in the form. Based on these defect rate distributions, the maintenance content of the equipment is determined.

More specifically, the costs of the pre-check and the post-check are calculated by using a computer or other suitable arithmetic means based on the result of the defect rate distribution, so that a continuous defect rate for each equipment can be calculated. You can check the inspection cost. At this time, the cost of the complete replacement of the device (complete failure) and the replacement of some parts (partial failure) differ as the degree of failure. Therefore, the degree of failure is classified into complete failure and partial failure from the inspection result, and the failure rate distribution of each is determined. As a result, an actual cost can be calculated.

The building side compares the result of estimating the failure rate of the equipment of the building with the actual result as described above, and determines whether the pre-check or the post-check is more economical or the type of the equipment. This makes it possible to judge how to incorporate both the pre-check and the post-check. Also, the inspector can objectively present appropriate maintenance contract contents.

As a method of estimating the defective rate, a Bayesian approach (a method based on Bayesian statistics) that can estimate the overall defective rate by performing a small number of sampling inspections several times is used.

According to the present invention, as described above, the results of the pre-inspection performed periodically regardless of the presence or absence of a failure and the results of the post-inspection performed at the time of occurrence of a failure on the equipment to be subjected to the maintenance and inspection are described below. , The costs of the pre-inspection and the post-inspection are calculated, these two costs can be objectively evaluated, and a reasonable and appropriate maintenance contract can be concluded.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A case where the method of the present invention is applied to maintenance and inspection of an automatic air-conditioning control device of a building will be described.

First, inspections of the equipment are periodically performed as "preliminary inspection", and the other inspections are defined as "post inspection".

The types of air conditioning automatic control equipment to be subjected to maintenance and inspection are very large, exceeding 2,000. Therefore, these devices are classified according to their characteristics. For example, DD
C (Direct Digital Controller), detector, operating device, regulator, valve, etc.

The inspection result of the "preliminary inspection" is determined in several types depending on the state of the device. When focusing on the device determined to be defective, the case where the device is completely replaced with the product and the case where the component of the device is replaced are completely replaced. There are cases where the costs are different, and the costs are naturally different between the two. Therefore, the former is distinguished as "complete failure" and the latter is distinguished as "partial failure". In other words, "preliminary inspection"
Judgment of defective products at the time is divided into "complete defect" and "partial defect". However, in the case of the "post-test", since only two types, "normal" and "defective", are determined, all the defects are regarded as "complete defects".

From the inspection results of the above-mentioned preliminary inspection and post-examination, the defect rate and the prior probability of the equipment in each case are calculated. Here, the "prior probability" refers to a prior (sample) probability distribution. For example, defect rate 3%
・ A prior probability of 10% means that as a result of inspection,
% Is 10% ".

The main procedure for implementing the method of the present invention is as follows. <Step 1> The equipment to be inspected is classified as described above. <Step 2> The failure rate (mean μ, standard deviation σ) of each classified device is calculated, and a failure rate distribution f (x, μ, σ) (normal distribution) is created. <Step 3> Failure rate distribution f (x, μ, σ) (normal distribution)
Is defined as n (square root of the number of defects) areas S ₁ ,..., S _i ,
..., it is divided into S _n.

The percent defective area S _i ranges, a _i-1 ≦ x <
a _i . However,

[0025]

(Equation 1) <Step 4> failure rate is defined by the following equation fp _i and prior probabilities rp
Calculate _i .

[0026]

(Equation 2) The above equation, the centroid of each area S _i and failure rate fp _i, each area S
Ratio when the total area of _i is normalized to 1 (area ratio)
The means that the prior probability rp _i. That is, the center of gravity of each region is set as a representative value of the defect rate, and the area ratio of the region is set as the prior probability at the defect rate.

In the case of "preliminary inspection", the failure rate is the sum of the complete failure rate and the partial failure rate. The complete failure rate and the partial failure rate are calculated as "complete failure distribution" and "partial failure rate" in each area. Distribution according to the area ratio.

According to the above procedure, the defect rate and the prior probability of the pre-inspection and the post-inspection for the devices classified as described above are calculated.

Next, the calculation of the inspection cost will be described.

As shown in Table 1 below, in the case of "preliminary inspection", an inspection cost (artificial cost) and a cost for a defective product (a product cost for a complete defect or a component cost for a partial defect) are required. That is, even if there is no faulty product, an inspection cost is required for one device, but no artificial cost (replacement cost) is required for the defective product. On the other hand, in the case of "post-test", there is no inspection cost, but defective products (all are treated as completely defective as described above)
Replacement costs (artificial costs) and product costs (goods costs).

[0031]

[Table 1] A certain defect rate fp (= complete defect rate fp _all + partial defect rate fp
The inspection cost (including both pre- and post-inspection) in _prt ) is calculated from the following equation.

[0032] _{M {kc + fp all · (} c pdt + c man) + fp prt · (c prt + c 'man)} ... (4) However, M is checking the number, kc inspection costs, c _pdt product cost, c
_prt is a part cost, c _man is an artificial cost at the time of product replacement, and c ′ _man is an artificial cost at the time of component replacement.

Next, assuming that the defect is only a complete defect and the cost of the pre-inspection and the cost of the post-inspection are equal, the following relationship is established.

[0034]

(Equation 3) Based on this relationship and the result of estimating the failure rate by the “Bayesian approach” described later, it is possible to set the costs of the preliminary inspection and the post-inspection in a well-balanced manner.

[0035]

1 and 2 are flowcharts showing a detailed procedure for implementing a method of the present invention using a computer.

First, the equipment is classified according to its characteristics based on the result of the inspection (ST1). This classification is executed by inputting data indicating which of the above-mentioned six types of equipment is to the computer. As the input method, keyboard operation by an operator, reading from a medium such as an FD (flexible disk) or a CD (compact disk) storing those data,
Transmission by a communication network or the like is used.

Next, the computer executes a process of determining whether or not the performed inspection is a "prior inspection" (ST).
2) In the case of “No”, the procedure of FIG. 2 is executed as “post-check” as indicated by the connector. For this,
I will explain later.

In the case of “preliminary inspection” in the above judgment, the state of each classified device is determined as “normal”,
It is divided into "partial failure" and "complete failure" (ST3). This classification is entered by a human into a computer.

The computer executes the following processing.
The failure rate (mean μ, standard deviation σ) of each classified device is calculated, and a failure rate distribution A is created according to a normal distribution (S
T4). The failure rate distribution A is divided into a complete failure distribution a1 and a partial failure distribution a2 (ST5). ± 3σ of failure rate distribution A
The (standard deviation) is divided into n (square root of the number of defects) areas (ST6). Using the center of gravity of each area as a representative value of the defect rate,
The ratio of the area of the region is set as the prior probability at the defect rate (ST7).

Next, the computer calculates the respective averages from the inspection costs, the product cost of completely defective products, and the component cost of partially defective products, which are input in advance (ST8). Then, M (number of inspections) x [kc (inspection cost) + fp _all (perfect defect rate) x c
_pdt by (product cost) + fp _prt (part failure rate) × c _prt (parts cost)] ... (6), to calculate the "pre-inspection costs" in each failure rate (ST9).

On the other hand, in the case of "post inspection", as shown in FIG. 2, the state of each classified device is divided into "normal" and "completely defective" according to the inspection result (ST21). This classification is also input to a computer by a human.

The computer executes the following processing.

The failure rate (mean μ, standard deviation σ) of each classified device is calculated, and a failure rate distribution B is created according to a normal distribution (ST22). This failure rate distribution B is only a complete failure distribution. Next, ± 3σ (standard deviation) of the failure rate distribution B is divided into n (square root of the number of failures) areas (S
T23). Then, the center of gravity of each area is set as a representative value of the defect rate, and the ratio of the area of the area is set as the prior probability at the defect rate (ST24).

Next, the computer calculates respective averages from the input product cost and artificial cost of each device (ST25). Then, the “post-check cost” at each defect rate is calculated by M (number of inspections) × fp _all (complete defect rate) × [ _cpdt (product cost) + c _man (artificial cost)] (ST26) ). As a result, as shown by the connector, FIG.
The next step (ST10) after ST9 can be executed.

Referring again to FIG. 1, the computer displays the "preliminary inspection cost" and the "post-inspection cost" calculated in the above procedure on a "defective rate versus cost graph" (ST1).
0). By referring to this graph, both the inspector and the building can easily understand the relationship between the defect rate and the inspection cost, and can objectively compare and evaluate the maintenance contract (especially the cost). Specifically, the cost can be compared by examining the “inspection cost” at the time of the preliminary inspection or the “artificial cost” at the time of the post-inspection.

Therefore, as a procedure after the display of the “defective rate versus cost graph”, the computer can execute the following processing. That is, when a person decides whether or not to change the "inspection cost" and inputs it to the computer, the computer judges the change (ST11), changes the inspection cost to the input value (ST12), and based on that. The preliminary inspection cost is calculated (ST9) and the graph is displayed (ST10). Also, when a human inputs a change in "artificial cost", the computer determines the change (ST13), changes the artificial cost to the input value (ST14), and based on that, as shown by the connector. After calculating the post-check cost of FIG. 2 (ST26), the graph display of FIG. 1 (ST10) is performed.

With the above, the procedure of FIGS. 1 and 2 is completed.
According to the present invention, a program for causing a computer to execute the above processing is recorded on a medium such as an FD or a CD, and when the method of the present invention is performed, the content of the medium is read by the computer. You may do so.

FIG. 3 shows “preliminary inspection” and “post-inspection” performed for each device classified into six in the above procedure.
3 shows the defect rate distribution based on the results of the above. Here, the failure rate distribution A of the preliminary inspection is the sum of the complete failure rate distribution a1 and the partial failure rate distribution a2. However, in the case of "DDC" of the above classification,
Since there was no partial defect at the time of the preliminary inspection and no defect at the time of the post inspection, only the complete defect rate a1 at the time of the preliminary inspection was obtained.

Apart from the above “DDC”, in the equipment of the classification except for the “other” of the classification, the post-mortem inspection has a higher average failure rate μ and a larger standard deviation σ. In addition, a preliminary inspection shows that the appearance of complete failures and partial failures, which are the breakdown of the failures, has characteristics depending on each device.

FIG. 4 shows the result of dividing the area of the defect rate distribution for the "operator" of the classification as an example of the area division of the defect rate distribution as described above (ST6 and ST23). Table 2 below shows the prior probabilities with respect to the defect rates obtained as the centers of gravity of these regions.

[0051]

[Table 2] FIG. 5 shows an example of the result (graph display of ST10) of calculating the pre-inspection cost (ST9) and the post-inspection cost (ST26) for the failure rate of the classified equipment (operator, detector, controller, valve). . It should be noted that "DDC" has no defect at the time of the post-examination, and that the equipment classified as "Other" has a large price variation, so that both are omitted.

In the pre-inspection, the defect rate is lower than in the post-inspection, so that the portion that cannot be obtained from the actual data is determined to be either a complete defect or a partial defect based on the previous defect tendency, and the inspection cost is reduced. Calculated.

In the post inspection, as described above, "complete failure"
Since only the defect rate is taken into account, the defect rate is proportional to the inspection cost.However, in the case of the pre-inspection, the percentage of complete defect and partial defect in each defect rate is different in addition to the inspection cost. High and low are shown. For example, "operator"
Therefore, the cost of the preliminary inspection was not much increased because only the partial defect was considered as a defect based on the tendency before that. Also, in the case of the "valve", the cost of the preliminary inspection and the post-inspection are equal at around 4% of the defect rate due to the relationship between the complete defect and the partial defect. Costs are reversing.

The arrow in FIG. 5 indicates the failure rate at which the "pre-check" cost and the "post-check" cost are equal, and it is determined whether the cost of the pre-check or the post-check is lower. To serve. If the defect rate is higher than this, the preliminary inspection is cheaper than the post inspection.

Finally, the "Bayesian approach" will be described.

According to Bayesian statistics, the “posterior probability” is a combination of the “prior probability” set before the experiment and the experiment result (sample information), and it is possible to use this to infer items related to the parameter. it can.

The “Bayesian approach” is defined as follows, and is applied to “calculation of expected failure rate”.

[0058] failure rate _{fp 1, ..., fp i,} ..., when considering the fp _z, respectively rp ₁ a prior probability for the failure rate,
…, Rp _i ,…, rp _z . However,

[0059]

(Equation 4) At this time, when there are r defective products in the n samples, the posterior probabilities op _i are defined as follows.

[0060]

(Equation 5) Here, joint probability: sp _i = rp _i · u _i (10) Likelihood: u _i = _n C _r (fp _i ) ^r (1−fp _i ) ^nr (11) Also, the expected failure rate at this time efp is represented by the following equation.

[0061]

(Equation 6) Further, the k-th expected failure rate efp _k can be _obtained by repeatedly calculating the posterior probability op _i for the initial prior probability rp _{i | 1} as the prior probability rp _{i | 2} . By utilizing the fact that this value goes in the convergence direction, the defect rate of the parameter can be estimated with a small total number of samplings. That is, if the total number of pieces (pieces) obtained by the number of pieces (pieces / times) × the number of times of sampling (times) converges with a number smaller than the population parameter and an accurate failure rate can be estimated, this method is meaningful.

Therefore, the "parameters" (pieces) are sampled by the "number of pieces" (pieces), and a "good" or "defective" judgment is made for each piece according to the "set defect rate" (actual failure at this time). Rate is referred to as the “appearance failure rate”). Expected failure rate
When the condition that the absolute deviation from efp one time before is within the "convergence determination width" continues for "the number of times of convergence determination", it is determined that efp has converged.

FIG. 6 shows, as a simulation sample, the convergence state of the “expected defect rate” and the number of defects that have appeared.

The absolute deviation of the expected failure rate is 1 from the tenth time.
It can be seen that it is within 0.1% continuously until the 1st, 12th, and 13th times, and converges to 3.1% as a result.

In the verification, 10 simulations were performed under the same conditions.

As an example, FIG. 7 shows the “operator” of the preliminary inspection.
For the defect rate and prior probability of, the parameter is 2,400,
The results of the average convergence failure rate and the average total number of samplings estimated by the simulation when the convergence determination width is changed with the number of convergence determinations being three are shown.

Regardless of the combination of the setting failure rate and the number of samples, the convergence judgment width increases and the average number of samples decreases, but there is no significant change from the convergence judgment width of about 0.3 to 0.5%. This means that the expected failure rate
Probably because it is around 0.3-0.5%. When the setting failure rate is 2%, the value often converges to the + side of the “appearance failure rate”, but when the setting failure rate is 4%, the opposite is true. This is thought to be related to the failure rate and how to obtain the initial prior probability.

Tables 3 and 4 show the average total number of samples and the average expected failure rate width excluding the convergence judgment width 0.1 (%) of each graph.

[0069]

[Table 3]

[0070]

[Table 4] The number of samples is small (24 pieces) and the setting failure rate is low (2
%), The average total number of samplings is the smallest (about 4% of the population), and an accurate failure rate is estimated. It is considered that the reason for this is that the smaller the number of samples to be extracted and the lower the defective setting rate, the smaller the variation in the number of defectives, and quickly and accurately reach the convergence value.

When the number of times of convergence (= the total number of samplings / the number of samplings) is calculated, there is no large difference of about 4 times in any combination.

In the above, regarding the evaluation of the pre-inspection and the post-inspection based on the defect rate, the classification of the defective product and the calculation of the inspection cost based on the distribution of the defect rate have been described by taking an example of an automatic control device for air conditioning. In addition, the maintenance content of various equipment can be appropriately and objectively determined by obtaining the defect rate distribution of the pre-inspection and the post-inspection.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a detailed procedure for performing the method of the present invention.

FIG. 2 is a flowchart showing a procedure branched from FIG. 1;

FIG. 3 is a graph showing a defect rate distribution based on the results of “pre-inspection” and “post-inspection” performed for each of the six types of equipment.

FIG. 4 is a diagram showing a result of dividing an area of a failure rate distribution with respect to an example of an operating device of a classified device.

FIG. 5 is a graph showing an example of a pre-inspection cost and a post-inspection cost with respect to a failure rate of classified devices (operating devices, detectors, controllers, and valves).

FIG. 6 is a graph showing the convergence state of the “expected failure rate” by the Bayesian approach and the number of failures that have appeared.

FIG. 7 is a graph showing the results of the average convergence failure rate and the average total number of samplings estimated by the simulation when the convergence determination width is changed in the case of the failure rate and the prior probability of the “operator” in the preliminary inspection.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Yukihiko Higashi 4-3-4 Shibaura, Minato-ku, Tokyo Yamatake Keiso Co., Ltd. (72) Inventor Hirofumi Miyoshi 4-3-4 Shibaura, Minato-ku, Tokyo Yamatake Instrumentation Co., Ltd. (72) Inventor Miyasaka Fusaka 2-12-19 Shibuya, Shibuya-ku, Tokyo Yamatake Ha Newel Co., Ltd. (72) Inventor Eizo Kinoshita 3-53 Niigaoka, Kani City, Gifu Prefecture

Claims (4)

[Claims] An inspection of equipment subject to maintenance inspection is divided into a pre-inspection that is performed periodically regardless of the presence or absence of a failure and a post-inspection performed when a failure occurs. From both results, a failure rate distribution by pre-inspection and a failure rate distribution by post-inspection of the equipment are obtained, and maintenance contents of the equipment are determined based on these two failure rate distributions. Method. 2. The method according to claim 1, wherein each of the two defect rate distributions is multiplied by a cost required for replacement of a defective product, thereby calculating the cost of each of the pre-check and the post-check. A method comprising: 3. The method according to claim 1, wherein the degree of failure of the equipment is divided into a complete failure for replacing the whole equipment and a partial failure for replacing parts, and in the case of the preliminary inspection, A method of calculating a sum of defect rate distributions of a complete defect and a partial defect, and in the case of the post-check, determining a defect rate distribution of only the complete defect. 4. A medium storing a program for causing a computer to execute a process of determining a failure rate distribution of equipment to be subjected to maintenance and checking and determining maintenance contents of the equipment based on the distribution. The program includes a process for determining whether the inspection of the equipment is a pre-inspection performed periodically regardless of the presence or absence of a failure or a post-inspection performed when a failure occurs, and both the pre-inspection and the post-inspection. From the results of the above, a process of obtaining a failure rate distribution by prior inspection of the equipment and a failure rate distribution by post-inspection, and each of the preliminary inspection and the post-inspection based on the two failure rate distributions and the cost required for replacement of defective products. And a process for calculating the cost of the equipment.