JP2023062423A - Degradation diagnosis device, method for diagnosing degradation, and degradation diagnosis program - Google Patents

Abstract

To detect a physical quantity accurately even when a temperature hysteresis is happening.SOLUTION: A degradation diagnosis device 1 includes: an acceleration sensor 10a for measuring an inclination amount of an infrastructure facility; a temperature sensor 10b for measuring the temperature in the position of the acceleration sensor 10a; and a controller 11 for regularly acquiring data of the correspondence between the inclination amount and the temperature. The controller 11 learns a temperature correction expression of the inclination amount on the basis of a plurality of pieces of correspondence data in a predetermined learning period of time, and corrects the inclination amount acquired in an operation period of time by the temperature correction expression.SELECTED DRAWING: Figure 1

Description

The present invention relates to a deterioration diagnosis device, a deterioration diagnosis method, and a deterioration diagnosis program.

In recent years, the aging of infrastructure facilities such as bridges, tunnels, and building structures has become a social problem, and various techniques have been researched and developed to check the state of deterioration over time and to take countermeasures. For example, as a method of detecting the state of deterioration of infrastructure equipment as described above, physical quantity applied to the infrastructure equipment is measured by physical quantity sensors such as acceleration sensors, strain sensors, or load sensors installed in the infrastructure equipment, and the physical quantity applied to the infrastructure equipment is measured. A method using changes in the physical quantity over a long period of several decades as an index is exemplified.

By the way, it is known that the measurement accuracy of the physical quantity sensor described above is greatly affected by the surrounding environmental temperature. many. For example, in the prior art disclosed in Patent Document 1, temperature correction of the tilt amount is performed by approximating the correspondence relationship between the tilt amount and the temperature with a curve of a first order or a second order or higher.

However, in the conventional technology as described above, since it is assumed that the relationship between the temperature shift of the tilt amount and the temperature is one-to-one, if a hysteresis occurs in the relationship between the two that is actually measured, , there arises a problem that the measurement accuracy of the tilt amount is lowered.

On the other hand, in the conventional technology disclosed in Patent Document 2, attention is paid to the fact that temperature hysteresis may occur in which the bias is different between when the temperature rises and when the temperature falls. , discloses a temperature compensation process in which the temperature hysteresis characteristics of a pair of acceleration sensors installed in opposite directions are offset by an addition process.

JP 2016-145709 A JP 2019-163955 A

However, the prior art of Patent Document 2 requires a pair of acceleration sensors for each axial direction, and measurement errors due to individual differences between the sensors cannot be resolved. occur. In particular, when detecting the deterioration state of infrastructure equipment, if the amount of inclination (inclination angle) is used as a physical quantity, it is required to keep the measurement error within about ±0.01°. Achieving measurement accuracy is difficult.

The present invention has been made in view of such problems, and aims to provide a deterioration diagnosis device, a deterioration diagnosis method, and a deterioration diagnosis method capable of detecting a physical quantity with high accuracy even when temperature hysteresis occurs. and to provide a deterioration diagnosis program.

In order to achieve the above object, the deterioration diagnosis apparatus of the present invention comprises a physical quantity sensor for measuring a physical quantity for diagnosing deterioration of infrastructure equipment, a temperature sensor for measuring a temperature at a position of the physical quantity sensor, the physical quantity and the a control device that periodically acquires temperature correspondence data, wherein the control device learns the temperature correction formula for the physical quantity based on a plurality of the correspondence data in a predetermined learning period, and acquires the temperature correction formula during the operation period. is corrected by the temperature correction formula.

Further, the deterioration diagnosis method of the present invention includes a data acquisition procedure for periodically acquiring correspondence data of physical quantities and temperatures for diagnosing deterioration of infrastructure equipment, and A learning procedure for learning a temperature correction formula for a physical quantity, and a correction procedure for correcting the physical quantity acquired during an operation period based on the temperature correction formula.

Further, the deterioration diagnosis program of the present invention includes a data acquisition procedure for periodically acquiring correspondence data of physical quantities and temperatures for diagnosing deterioration of the infrastructure equipment, and a predetermined A learning procedure for learning a temperature correction formula for the physical quantity based on the plurality of corresponding data in the learning period, and a correction procedure for correcting the physical quantity acquired during the operation period based on the temperature correction formula are executed. .

According to the degradation diagnostic device, degradation diagnostic method, and degradation diagnostic program according to the present invention, physical quantities can be detected with high accuracy even when temperature hysteresis occurs.

It is a block diagram which shows the structure of a degradation diagnostic apparatus. It is a waveform representing the change in the amount of inclination and the temperature before temperature correction. 4 is a flow chart showing a control procedure during a learning period of a deterioration diagnosis method; 4 is a flow chart showing a control procedure during the operation period of the deterioration diagnosis method; It is a waveform showing changes in the temperature and the amount of tilt before and after the temperature correction during the learning period and the operating period.

Hereinafter, embodiments of the invention will be described in detail with reference to the drawings. It should be noted that the present invention is not limited to the contents described below, and can be arbitrarily changed and implemented without changing the gist of the present invention. In addition, all of the drawings used to describe the embodiments schematically show constituent members, and are partially emphasized, enlarged, reduced, or omitted for the purpose of deepening understanding. It may not represent the scale, shape, etc. accurately.

FIG. 1 is a block diagram showing the configuration of the deterioration diagnosis device 1. As shown in FIG. A deterioration diagnosis device 1 is installed in infrastructure facilities such as bridges, tunnels, and building structures, and continuously detects slight changes in physical quantities that occur over several years to several decades in order to grasp the deterioration state of the infrastructure facilities. It is a remote monitoring device. Here, an embodiment using a tilt amount as a physical quantity for diagnosing deterioration of infrastructure equipment is exemplified, but strain or load applied to the infrastructure equipment may be used.

Further, the deterioration diagnosis device 1 in this embodiment is connected to the operation terminal 3 via the cloud server 2 by two-way communication. The cloud server 2 stores various data received from the deterioration diagnosis device 1 and manages the deterioration diagnosis device 1 by controlling the deterioration diagnosis device 1 based on control signals from the operation terminal 3 . The operation terminal 3 is composed of, for example, a general-purpose computer, and by accessing the cloud server 2, the user can refer to various data of the deterioration diagnosis device 1, and by receiving user operations, data can be obtained via the cloud server 2. to control the deterioration diagnostic device 1 .

The deterioration diagnosis device 1 includes a sensor section 10, a control device 11, a storage device 12, a communication section 13, a battery 14, a sensor power supply section 15, and a communication power supply section 16, which are accommodated in a waterproof housing.

The sensor unit 10 is a MEMS sensor (Micro Electro Mechanical Systems) including an acceleration sensor 10a and a temperature sensor 10b as "physical quantity sensors". The acceleration sensor 10a is a triaxial inclinometer that measures the amount of inclination of the infrastructure facility in which the deterioration diagnosis device 1 is installed. The temperature sensor 10b measures the temperature at the position of the acceleration sensor 10a. Of the X-, Y-, and Z-axes, only the tilt amount of the X-axis will be described below, and detailed description of the tilt amounts of the Y-axis and Z-axis, which are similarly measured and calculated, will be omitted. When measuring the strain and load applied to infrastructure equipment, strain gauges and load gauges are used as "physical quantity sensors."

The control device 11 is composed of a known microcomputer control circuit and includes an internal timer called RTC (Real-Time Clock) to periodically transmit tilt amount and temperature data acquisition commands to the sensor section 10 . The control device 11 receives the corresponding data consisting of the tilt amount and the temperature acquired by the sensor unit 10 , saves it in the storage device 12 together with the measurement time, and also transmits it to the cloud server 2 via the communication unit 13 . Furthermore, the control device 11 performs calculations for learning a temperature correction formula with corresponding data measured in a prior learning period in order to estimate the temperature shift amount included in the measured value of the tilt amount, as will be described later in detail. I do.

Here, in this embodiment, the control device 11 transmits a data acquisition command to the sensor unit 10 every hour. At this time, the sensor unit 10 measures the tilt amount multiple times with the acceleration sensor 10 a , measures the temperature at each time with the temperature sensor 10 b , and returns the data to the control device 11 . Then, the control device 11 can eliminate sudden disturbances such as vibrations and variations in measurement by calculating the average of the tilt amounts measured a plurality of times. Then, the corresponding data for each hour is composed of the averaged value of the tilt amount and the measured temperature. Note that the averaging calculation is not an essential component, and various known noise processing may be employed.

The storage device 12 consists of a known non-volatile memory, receives and stores measured corresponding data and various parameters calculated with respect to the temperature correction formula through serial communication with the control device 11, and stores such information as necessary. is read out from the control device 11.

The communication unit 13 includes a known LPWA module (Low Power Wide Area) and a planar antenna, and performs wireless communication with the cloud server 2 . It should be noted that an external rod antenna may be used in place of the planar antenna inside the housing of the deterioration diagnosis device 1 .

The battery 14 is, for example, a lithium battery with a rated voltage of about 3 V, and supplies power to the control device 11 , the sensor section 10 and the communication section 13 . The control device 11 monitors the remaining power of the battery 14 and notifies the operation terminal 3 via the communication unit 13 and the cloud server 2 when the battery needs to be replaced.

The sensor power supply unit 15 and the communication power supply unit 16 are switches composed of MOSFETs (Metal-Oxide-Semiconductor Field-Effect Transistors). The power supply to 13 is controlled on/off.

With the above-described configuration, the deterioration diagnosis device 1 can be used as an infrastructure facility at a low cost and with low power consumption by means of a communication module that suppresses the data transmission capacity and an energy-saving design that supplies power from the battery 14 to each component only when necessary. can be monitored.

Next, the subject of the present invention will be described in detail. FIG. 2 shows waveforms representing changes in the amount of tilt and temperature before temperature correction. More specifically, FIG. 2 shows waveforms of the tilt amount and the temperature for six days measured by the acceleration sensor 10a and the temperature sensor 10b as described above. Although the average value of the tilt amount is not 0°, this depends on the installation angle of the deterioration diagnostic device 1 .

As shown in FIG. 2, the measured value of the slope amount before temperature correction has a strong correlation with the diurnal temperature change, and includes a temperature shift of 0.1° or more due to daily temperature fluctuations. Also, the change in the tilt amount is temporally delayed with respect to the change in temperature, and temperature hysteresis can be confirmed in which the relation between the tilt amount and the temperature does not correspond one-to-one. For this reason, for example, in the conventional correction method in which the temperature characteristics of the tilt amount are prepared in advance and the temperature is corrected by referring to the temperature at the time of measuring the tilt amount, the error is large and the tilt amount cannot be detected with high accuracy. Gone.

Therefore, the deterioration diagnosis device 1 of the present invention learns a temperature correction formula that takes into account the temperature hysteresis of the tilt amount during the preliminary learning period by executing the deterioration diagnosis program shown below in the control device 11, and during the operation period, A tilt amount is detected with high accuracy by a deterioration diagnosis method for correcting the obtained tilt amount based on the temperature correction formula. Here, the deterioration diagnosis method will be described by dividing it into a preliminary learning period and an operation period.

FIG. 3 is a flow chart showing the control procedure during the learning period of the deterioration diagnosis method. After the deterioration diagnosis device 1 is installed in the infrastructure equipment and the communication environment with the cloud server 2 and the operation terminal 3 is established as shown in FIG. 1, the procedure shown in FIG. 3 is started based on the instruction to start learning from the user. do.

When the learning period starts, the control device 11 starts a data acquisition procedure for periodically acquiring data corresponding to the tilt amount and the temperature of the infrastructure equipment by the sensor unit 10 (step S1). In the data acquisition procedure, slope and temperature are measured periodically at hourly intervals, as described above. Note that the data acquisition interval may be changed by an instruction from the operation terminal 3 .

Further, the control device 11 performs data acquisition until the correspondence data of the tilt amount and the temperature for one day is acquired (No in step S2), and when the correspondence data for 24 hours is accumulated (Yes in step S2). ), and the temperature correction formula is learned using the corresponding data (step S3, learning procedure).

As preprocessing for learning the temperature correction formula, the control device 11 calculates the average value of the obtained correspondence data for 24 hours, and uses the difference between the correspondence data and the average value to perform subsequent calculations. . More specifically, with respect to the tilt amount, the difference ΔX=x−X _AVE , where X _AVE is the average value of the measured tilt amounts x, and with respect to the temperature, T _AVE is the average value of the measured temperature t. Then, by performing the calculation using the difference ΔT=t−T _AVE , it is possible to prevent the cancellation error in the calculation of the control device 11 .

Further, the temperature correction formula in the present embodiment is modeled as a formula expressing the temperature shift amount of the latest tilt amount by a weighted linear sum of a plurality of temperatures acquired in the most recent predetermined time. More specifically, if the temperature n hours ago is _ΔTn for an integer n, and the weight parameter is a=[a ₀ , a _{1 ,} . The amount x _shift is represented by the following formula.
x _shift =f(ΔT _n , a)=a ₀ ΔT ₀ +a ₁ ΔT ₁ +...+a _n ΔT _n Expression (1)

Here, the integer n in equation (1) is a predetermined time assumed as a delay time of the change in the tilt amount with respect to the temperature change, and is arbitrarily set in advance. However, the predetermined time can be set by the above-described delay time of the temperature hysteresis. may be set as In this way, by estimating the delay time of the change in the slope amount with respect to the temperature change during the learning period and setting the predetermined time according to the delay time, the machine learning described later can be performed while covering the influence time of the temperature hysteresis. Convergence time can be shortened.

During the learning period, since the change in the true tilt amount can be assumed to be 0, the control device 11 sets the weighting parameter a such that the temperature shift amount x _shift becomes 0 in the obtained corresponding data. Calculated by machine learning based on That is, in order to calculate the value a _FIX of the weighting parameter a such that the error E becomes E=Σ(x _shift ) ² →0, the update formula a _NEW =a _OLD -η·dE/da is used for iterative optimization calculations.

The initial value of the weighting parameter a can be obtained from data acquired before installation of the deterioration diagnosis device 1. It is stored in the control device 11 . Also, the learning coefficient η can be arbitrarily set in advance, and is set to η=0.00001 here. Based on these parameters, the first calculation yields E = 0.00456, dE/da = [0.789, -0.805, -0.803, 0.817]. -0.00266, -0.00275, 0.00298]. By repeating the same calculation 1000 times, the error E converged from 0.00456 to 0.00001591, and the value a _FIX =[0.000167, -0.000143, -0.000177, 0.000181] at the time of convergence was obtained. As a result, the controller 11 can obtain a temperature correction formula as the temperature shift amount x _shift =f(ΔT _n , a _FIX ) of the tilt amount.

Further, the control device 11 confirms whether or not there is an instruction to end the learning period from the operation terminal 3 (step S4). Then, while there is no instruction to end the learning period, the procedure of steps S2 and S3 is repeated, and the learning of the temperature correction formula is updated based on the corresponding data acquired during this period. Here, in this embodiment, a four-day learning period is provided. It should be noted that the learning period does not necessarily have to be based on an end instruction from the operation terminal 3, and may end at an arbitrarily set period in advance.

Then, when the learning period ends (Yes in step S4), the control device 11 once terminates the periodical acquisition of correspondence data (step S5). In addition, the control device 11 saves these learning results in the storage device 12, transmits them to the cloud server 2 via the communication unit 13 (step S6), and terminates the pre-learning program.

Thereby, the control device 11 can learn the temperature correction formula using the corresponding data in the previous learning period. Here, the control device 11 can collectively learn the temperature correction formula using all the corresponding data acquired during the four days of the learning period. is divided into a plurality of calculation periods for each day, and the learning result of the temperature correction formula is updated for each calculation period. As a result, the calculation load of the control device 11 and the communication load of the communication unit 13 can be distributed temporally.

Next, the operation of the control device 11 during operation will be described. FIG. 4 is a flow chart showing a control procedure during operation of the deterioration diagnosis method. After the learning period ends, the deterioration diagnosis device 1 starts the procedure shown in FIG. 4 based on the operation start command from the user.

When the operation period starts, the control device 11 periodically (for example, every hour) acquires corresponding data similarly to the learning period (step S7, data acquisition procedure). In addition, the control device 11 updates the learned temperature correction formula x _shift =f(ΔT _n , a _FIX ) with respect to the acquired data after n hours (4 hours in this embodiment) from the start of the operation period. A temperature shift amount obtained by substituting temperatures for n hours is calculated and subtracted from the measured tilt amount (correction procedure). As a result, the control device 11 can perform temperature correction with temperature hysteresis taken into consideration for the obtained value of the tilt amount (step S8).

When the acquired tilt amount is temperature-corrected, the control device 11 saves the corresponding data acquired in step S7 and the corrected tilt amount in the storage device 12 as necessary, and transmits the corrected tilt amount via the communication unit 13. and transmits it to the cloud server 2 (step S9). As a result, the cloud server 2 can monitor the deterioration state of the target infrastructure equipment based on the tilt amount after the temperature correction, which is periodically received from the deterioration diagnosis device 1 .

Although not an essential configuration for the present invention, the control device 11 may update the learning result of the temperature correction formula with corresponding data acquired during a certain period from the start of the operation period. More specifically, the control device 11 determines whether or not a certain period of time (for example, one month) has elapsed since the start of operation, during which it can be assumed that the tilt amount has not changed (step S10). Based on the corresponding data acquired until the period elapses (No in step S10), the above-described temperature correction formula may be updated (step S11). Then, when the certain period of time has passed (Yes in step S10), the control device 11 finishes updating the learning result of the temperature correction formula.

Further, the control device 11 may determine whether the value of the tilt amount corrected for temperature is equal to or greater than a threshold arbitrarily set in advance (step S12). In this case, the control device 11 continues to monitor the tilt amount for several years to several decades until the tilt amount of the infrastructure becomes equal to or greater than the threshold (No in step S12).

On the other hand, when it is determined that the tilt amount after correction is equal to or greater than the threshold value (Yes in step S12), a tilt detection warning is sent to the operation terminal 3 via the communication unit 13 and the cloud server 2. (step S13), and the program may be terminated. Alternatively, monitoring may be continued even after the warning is issued until there is an operation period end command from the operation terminal 3 .

Next, effects of the present invention will be described. FIG. 5 shows waveforms representing changes in the temperature and the amount of tilt before and after the temperature correction during the learning period and the operating period. Here, the corresponding data is acquired over nine days, and the temperature correction formula learned in the first four days of the learning period is used to generate a waveform obtained by temperature-compensating the tilt amount acquired in the remaining five days of operation. It is indicated by a solid line. The temperature data for nine days is indicated by a broken line, and the change in the amount of inclination before temperature correction is indicated by a dashed line for reference.

Here, the average value X _AVE of the inclination amount and the average value T _AVE of the temperature used in calculating the amount of inclination after temperature correction are x and T for the past 96 hours (24H×4 days) from the time to be corrected. was calculated as X _AVE =−0.076 and T _AVE =9.646 using Moreover, since the temperatures in the most recent four hours were ΔT=[−9.81, −10.76, −11.71, −11.65], the temperature shift amount x _shift is calculated as follows.
x _shift = f(ΔTn, aFIX)
=-9.81*0.000167-10.76*(-0.000143)-11.71*(-0.000177)-11.65*0.000181=-0.000154
Therefore, the first data of the tilt amount after temperature correction in FIG. 5 is calculated as x=X _AVE +x _shift =-0.0762. Then, by repeating the same calculation, the waveform of "amount of tilt after correction" in FIG. 5 is obtained.

As can be seen in FIG. 5, the tilt amount before the temperature correction is subject to the temperature shift in accordance with the temperature change, resulting in diurnal fluctuations. On the other hand, the tilt amount corrected for temperature by the deterioration diagnosis method according to the present invention hardly changed during the learning period, and the difference between the maximum value and the minimum value of the tilt amount in FIG. 5 was 0.006981. The measurement error is suppressed within ±0.01°. That is, according to the degradation diagnosis method, since the influence of the temperature shift is within ±0.01°, it is possible to detect a slight change in the inclination amount of the infrastructure equipment that occurs over several years to several decades.

As described above, the deterioration diagnosis apparatus 1 according to the present invention acquires a plurality of correspondence data of physical quantities and temperatures of infrastructure facilities in time series, and learns a temperature correction formula based on the correspondence data. The amount of temperature shift superimposed on the value can be estimated corresponding to time-series changes. As a result, the deterioration diagnosis device 1 according to the present invention can detect physical quantities with high accuracy even when temperature hysteresis occurs.

<Embodiment of the present invention>
A deterioration diagnosis device according to a first aspect of the present invention comprises a physical quantity sensor for measuring a physical quantity for diagnosing deterioration of infrastructure equipment, a temperature sensor for measuring a temperature at a position of the physical quantity sensor, the physical quantity and the temperature and a control device that periodically acquires the correspondence data of the physical quantity, the control device learns the temperature correction formula for the physical quantity based on the plurality of the correspondence data in a predetermined learning period, and is acquired during the operation period The physical quantity is corrected by the temperature correction formula.

According to the degradation diagnosis device according to the first aspect, a plurality of correspondence data of physical quantities and temperatures of infrastructure equipment are obtained in time series, and a temperature correction formula is learned based on the correspondence data. The superimposed temperature shift amount can be estimated corresponding to the time-series change, and the physical quantity can be detected with high accuracy even when temperature hysteresis occurs.

In the deterioration diagnosis device according to a second aspect of the present invention, in the above-described first aspect, the temperature correction formula calculates the latest temperature shift amount of the physical quantity from the plurality of temperatures acquired in the most recent predetermined time. It is an expression expressed by a weighted linear sum by

According to the deterioration diagnosis device according to the second aspect, the temperature correction formula efficiently expresses the latest temperature change that affects the latest physical quantity measured by a low-order polynomial. The amount of calculation can be suppressed.

A deterioration diagnosis device according to a third aspect of the present invention is the second aspect described above, wherein the control device estimates a delay time of the change in the physical quantity with respect to the change in the temperature in the learning period, The predetermined time is set according to time.

According to the deterioration diagnosis device according to the third aspect, it is possible to shorten the convergence time of iterative optimization calculations in learning the temperature correction formula while constructing the temperature correction formula that can cover the influence time of the temperature hysteresis.

A deterioration diagnosis device according to a fourth aspect of the present invention is the degradation diagnostic device according to any one of the first to third aspects described above, wherein the control device divides the learning period into a plurality of calculation periods, and to update the learning result of the temperature correction formula.

According to the deterioration diagnosis device according to the fourth aspect, it is possible to temporally disperse the computation load of the control device and the communication load in transmitting the computation result.

A deterioration diagnosis device according to a fifth aspect of the present invention is the deterioration diagnosis device according to the fourth aspect described above, wherein the control device calculates an average value of the corresponding data for each calculation period, and The temperature correction formula is learned using the difference between .

According to the deterioration diagnosis device according to the fifth aspect, by using the measured corresponding data as the difference from the average value for learning the temperature correction formula, even when handling minute numerical values, the calculation of the control device It is possible to prevent the error of cancellation in

A deterioration diagnosis device according to a sixth aspect of the present invention is the degradation diagnosis device according to any one of the first to fifth aspects described above, wherein the control device detects the temperature according to the corresponding data acquired during a certain period from the start of the operation period. Update the learning result of the correction formula.

According to the deterioration diagnosis device according to the sixth aspect, the accuracy of temperature correction can be further improved by performing reinforcement learning of the temperature correction formula using the corresponding data acquired during the operation period.

A deterioration diagnosing method according to a seventh aspect of the present invention comprises a data acquisition procedure for periodically acquiring correspondence data of physical quantities and temperatures for diagnosing deterioration of infrastructure equipment; and a correction procedure for correcting the physical quantity obtained during operation based on the temperature correction formula.

According to the degradation diagnosis method according to the seventh aspect, a plurality of correspondence data of physical quantities and temperatures of infrastructure equipment are obtained in time series, and a temperature correction formula is learned based on the correspondence data. The superimposed temperature shift amount can be estimated corresponding to the time-series change, and the physical quantity can be detected with high accuracy even when temperature hysteresis occurs.

A degradation diagnosis method according to an eighth aspect of the present invention is, in the above-described seventh aspect, wherein the temperature correction formula calculates the latest temperature shift amount of the physical quantity from a plurality of the temperatures obtained at the most recent predetermined time. It is an expression expressed by a weighted linear sum by

According to the deterioration diagnosis method according to the eighth aspect, the temperature correction formula efficiently expresses the latest temperature change that affects the latest physical quantity measured by a low-order polynomial. can be suppressed.

A deterioration diagnosis method according to a ninth aspect of the present invention is the eighth aspect described above, in which, in the learning period, a delay time of change in the physical quantity with respect to the change in temperature is estimated, and according to the delay time, The predetermined time is set.

According to the deterioration diagnosis method according to the ninth aspect, it is possible to shorten the convergence time of iterative optimization calculations in learning the temperature correction formula while constructing the temperature correction formula that can cover the influence time of the temperature hysteresis.

A deterioration diagnosis method according to a tenth aspect of the present invention is, in any one of the seventh to ninth aspects described above, wherein the learning period is divided into a plurality of calculation periods, and the temperature correction formula is calculated for each calculation period. update the learning results of

According to the deterioration diagnosis method according to the tenth aspect, it is possible to temporally disperse the computation load in learning the temperature correction formula and the communication load in transmitting the computation result.

A deterioration diagnosis method according to an eleventh aspect of the present invention is, in the above-described tenth aspect, calculating an average value of the corresponding data for each calculation period, and using a difference between the corresponding data and the average value. to learn the temperature correction formula.

According to the deterioration diagnosis method according to the eleventh aspect, by using the measured corresponding data as the difference from the average value for learning the temperature correction formula, even when handling minute numerical values, the calculation of the control device It is possible to prevent the error of cancellation in

A deterioration diagnosis method according to a twelfth aspect of the present invention is, in any one of the seventh to eleventh aspects described above, a learning result of the temperature correction formula based on the corresponding data acquired in a certain period from the start of the operation period. to update.

According to the deterioration diagnosis method according to the twelfth aspect, the accuracy of temperature correction can be further improved by performing reinforcement learning of the temperature correction formula using corresponding data acquired during the operation period.

A deterioration diagnosis program according to a thirteenth aspect of the present invention is a data acquisition procedure for periodically acquiring correspondence data between physical quantities and temperatures for diagnosing deterioration of infrastructure equipment in a control device for diagnosing deterioration of infrastructure equipment. a learning procedure for learning the temperature correction formula for the physical quantity based on the plurality of corresponding data in a predetermined learning period; and a correction procedure for correcting the physical quantity acquired during the operation period based on the temperature correction formula. , is executed.

According to the degradation diagnosis program according to the thirteenth aspect, a plurality of correspondence data of the physical quantity and the temperature of the infrastructure equipment are obtained in time series, and the temperature correction formula is learned based on the correspondence data. The superimposed temperature shift amount can be estimated corresponding to the time-series change, and the physical quantity can be detected with high accuracy even when temperature hysteresis occurs.

A deterioration diagnosis program according to a fourteenth aspect of the present invention is the above-described thirteenth aspect, wherein the temperature correction formula calculates the latest temperature shift amount of the physical quantity from a plurality of the temperatures acquired in the most recent predetermined time. It is an expression expressed by a weighted linear sum by

According to the deterioration diagnosis program according to the fourteenth aspect, since the temperature correction formula efficiently expresses the latest temperature change that affects the latest physical quantity measured by a low-order polynomial, the control device The amount of calculation can be suppressed.

A deterioration diagnosis program according to a fifteenth aspect of the present invention, in the above-described fourteenth aspect, estimates a delay time of change in the physical quantity with respect to the change in temperature in the learning period, and The predetermined time is set.

According to the deterioration diagnosis program according to the fifteenth aspect, it is possible to shorten the convergence time of the iterative optimization calculation in learning the temperature correction formula while constructing the temperature correction formula that can cover the influence time of the temperature hysteresis.

A deterioration diagnosis program according to a 16th aspect of the present invention is, in any one of the 13th to 15th aspects described above, wherein the learning period is divided into a plurality of calculation periods, and the temperature correction formula is calculated for each calculation period. update the learning results of

According to the deterioration diagnosis program according to the sixteenth aspect, it is possible to temporally disperse the computation load in learning the temperature correction formula and the communication load in transmitting the computation result.

A deterioration diagnosis program according to a seventeenth aspect of the present invention, in the sixteenth aspect described above, calculates an average value of the correspondence data for each calculation period, and uses a difference between the correspondence data and the average value. to learn the temperature correction formula.

According to the deterioration diagnosis program according to the seventeenth aspect, by using the measured corresponding data as the difference from the average value for learning the temperature correction formula, even when handling minute numerical values, the calculation of the control device It is possible to prevent the error of cancellation in

An eighteenth aspect of the present invention is a deterioration diagnosis program according to any one of the thirteenth to seventeenth aspects described above, wherein the learning result of the temperature correction formula is obtained from the corresponding data acquired in a certain period from the start of the operation period. to update.

According to the deterioration diagnosis program according to the eighteenth aspect, the accuracy of temperature correction can be further improved by performing reinforcement learning of the temperature correction formula using the corresponding data acquired during the operation period.

Reference Signs List 1 deterioration diagnosis device 2 cloud server 3 operation terminal 10 sensor unit 10a acceleration sensor 10b temperature sensor 11 control device 12 storage device 13 communication unit 14 battery 15 sensor power supply unit 16 communication power supply unit

Claims (18)

a physical quantity sensor that measures a physical quantity for diagnosing deterioration of infrastructure equipment;
a temperature sensor that measures the temperature at the position of the physical quantity sensor;
a control device that periodically acquires corresponding data of the physical quantity and the temperature;
The control device learns a temperature correction formula for the physical quantity based on the plurality of corresponding data in a predetermined learning period, and corrects the physical quantity acquired during the operation period with the temperature correction formula. 2. The deterioration diagnosis device according to claim 1, wherein said temperature correction formula is a formula expressing the latest temperature shift amount of said physical quantity by a weighted linear sum of said plurality of temperatures acquired in the most recent predetermined time. 3. The deterioration diagnosis device according to claim 2, wherein said control device estimates a delay time of change in said physical quantity with respect to said change in temperature during said learning period, and sets said predetermined time according to said delay time. 4. The deterioration diagnosis device according to claim 1, wherein said control device divides said learning period into a plurality of calculation periods, and updates the learning result of said temperature correction formula for each of said calculation periods. 5. The deterioration diagnosis according to claim 4, wherein the control device calculates an average value of the corresponding data for each calculation period, and learns the temperature correction formula using a difference between the corresponding data and the average value. Device. 6. The deterioration diagnosis device according to claim 1, wherein said control device updates the learning result of said temperature correction formula with said corresponding data acquired during a fixed period from the start of said operation period. a data acquisition procedure for periodically acquiring physical quantity and temperature correspondence data for diagnosing deterioration of infrastructure equipment;
a learning procedure for learning a temperature correction formula for the physical quantity based on the plurality of corresponding data in a predetermined learning period;
and a correction procedure for correcting the physical quantity acquired during operation based on the temperature correction formula. 8. The deterioration diagnosis method according to claim 7, wherein said temperature correction formula is a formula that expresses the latest temperature shift amount of said physical quantity by a weighted linear sum of said plurality of temperatures acquired in the most recent predetermined time. 9. The deterioration diagnosis method according to claim 8, wherein in said learning period, a delay time of change in said physical quantity with respect to said change in temperature is estimated, and said predetermined time is set according to said delay time. 10. The deterioration diagnosis method according to claim 7, wherein said learning period is divided into a plurality of calculation periods, and the learning result of said temperature correction formula is updated for each of said calculation periods. 11. The deterioration diagnosis method according to claim 10, wherein an average value of said correspondence data is calculated for each said calculation period, and said temperature correction formula is learned using a difference between said correspondence data and said average value. 12. The deterioration diagnosis method according to any one of claims 7 to 11, wherein the learning result of the temperature correction formula is updated with the corresponding data acquired during a certain period from the start of the operation period.
For the control device that diagnoses the deterioration of infrastructure equipment,
a data acquisition procedure for periodically acquiring physical quantity and temperature correspondence data for diagnosing deterioration of the infrastructure;
a learning procedure for learning a temperature correction formula for the physical quantity based on the plurality of corresponding data in a predetermined learning period;
and a correction procedure for correcting the physical quantity obtained during operation based on the temperature correction formula. 14. The degradation diagnostic program according to claim 13, wherein said temperature correction formula is a formula expressing the latest temperature shift amount of said physical quantity by a weighted linear sum of said plurality of temperatures obtained in the most recent predetermined time. 15. The degradation diagnostic program according to claim 14, wherein in said learning period, a delay time of change in said physical quantity with respect to said change in temperature is estimated, and said predetermined time is set according to said delay time. 16. The degradation diagnostic program according to any one of claims 13 to 15, wherein said learning period is divided into a plurality of calculation periods, and the learning result of said temperature correction formula is updated for each of said calculation periods. 17. The degradation diagnosis program according to claim 16, wherein an average value of said correspondence data is calculated for each said calculation period, and said temperature correction formula is learned using a difference between said correspondence data and said average value. 18. The deterioration diagnosis program according to any one of claims 13 to 17, wherein the learning result of the temperature correction formula is updated with the corresponding data acquired during a certain period from the start of the operating period.

Images