JP4665533B2 - Method and program for extracting candidate abnormal parts in structure - Google Patents

Description

  The present invention relates to a method and program for extracting candidates for abnormal points in a structure, and in particular, is a technology for detecting abnormal points in cavities, junkers, and other structures in concrete structures, roads, and tunnels. The present invention relates to an abnormal part candidate extraction method and program for detecting an internal abnormality that cannot be diagnosed and reproducing an image.

  A method using a ground penetrating radar is known as a non-destructive method for inspecting anomalies inside concrete. An inspection device using a ground penetrating radar uses a pair of a transmitting antenna element and a receiving antenna element to radiate a radio wave of a predetermined frequency from the concrete surface and detect an abnormality by capturing a reflected wave from an internal abnormal point. It is a mechanism. In order to grasp the abnormality inside the concrete three-dimensionally using a pair of transmitting and receiving antenna elements, it is necessary to measure along the lattice points with a ground penetrating radar. However, a great deal of time is required for the measurement work along the grid points, and this method is generally not practical. As a solution to this problem, there has been proposed a multi-path three-dimensional imaging radar apparatus that receives an reflected wave by an arbitrary combination of transmission / reception antenna elements using an array antenna in which a plurality of transmission / reception antenna elements are arranged (patent) Reference 1).

A conventional method for diagnosing the presence or absence of abnormal locations using a three-dimensional imaging radar device is described below. Moreover, it shows in a flow format in FIG.
1. Data is measured (step S10), and the measured data is divided into sizes that can be processed by the computer (step S12).
(In an actual computer for image reproduction processing, the traversing length of measurement data that can be image-reproduced at a time is 2.5 m. Due to the problem of computer resources, image reproduction processing of a wide range of measurement data cannot be performed at once. .)
2. Image reproduction processing is performed on the divided data to obtain a three-dimensional transmission image (step S14).
3. The operator sees the obtained transmission image and visually diagnoses the presence / absence of an abnormal part (step S16).
4). If step S16 of this visual inspection is completed, it is checked whether or not there is data remaining (step S18), and if there is any data, the process returns to step S16 to perform processing for diagnosing the next transmission image (step S18).
5. Such operations are repeated from step S16 to S18 until there is no more divided measurement data.

  With such a multi-pass method, it is possible to grasp abnormalities inside the concrete three-dimensionally while traversing, and also to detect deep layer defects hidden behind obstacles on the surface of the structure There is.

However, if the above procedure is performed, while the amount of measurement data is small, the procedure can be used to diagnose the presence of an abnormal part. However, as the amount of measurement data increases, a computer obtains a transmission image. Regardless of the calculation time until and the presence or absence of abnormal parts, the operator's effort to diagnose all transmitted images cannot be ignored and becomes increasingly impractical.
JP 2002-323459 A

  The present invention is intended for diagnosis of presence / absence of an abnormal part using a 3D imaging radar apparatus, and obtains a transmission image by a computer by performing image reproduction processing only on measurement data that may contain the abnormal part. It is intended to provide a method and a program for automatically extracting candidate candidates for abnormal points for the purpose of reducing the calculation time until the operator, reducing the labor of diagnosing the transmitted image by the operator, and reducing the time from measurement to output of the diagnosis result.

  In order to achieve the above object, a method for extracting an abnormal part candidate inside a structure according to the present invention is a method for detecting an abnormal part by inspecting the inside of the structure with a ground penetrating radar. Identify candidate areas, aggregate / evaluate reflected signal intensities contained in the obtained abnormal spot candidate areas, rank the abnormal spot candidate areas, and perform image reproduction processing from the measurement data of the higher places It is characterized by.

  The measurement data is compressed by using an array antenna in which a plurality of transmission / reception antenna elements are arranged, detecting data by a multipath system that receives a reflected wave with an arbitrary combination of transmission / reception antenna elements, and then detecting a plurality of data from the detected multipath data. It is possible to extract single path data from the pair of transmitting and receiving antenna elements and perform data compression by adding signal strength.

  The abnormal part is identified by extracting a normal part candidate area by removing a normal part from the reflection intensity distribution with a threshold value, and using the abnormal part candidate area and the reflection intensity included therein, an evaluation value indicating the degree of abnormality is obtained. Then, the position of the abnormal part candidate area is ranked by the evaluation value and specified.

The image reproduction processing from the measurement data can be performed based on the multipath data in order from the top of the abnormal part candidate areas ranked based on the single path data.
If an abnormal location is not found at a certain place, image reproduction processing and abnormal location diagnosis should not be performed for subsequent data.

The abnormal part candidate extraction program inside the structure according to the present invention is a program for detecting an abnormal part by performing an internal inspection of the structure by a ground penetrating radar, measuring multipath data from an array antenna, and the measurement calculating a data compression step, extracting abnormal part candidate region from the data the compressed, the evaluation value counts the reflected signal intensity included in the extracted abnormal point candidate area for compressing the multipath data was a step of, the abnormal point ranking step of ranking the anomaly candidate region based on the evaluation value described above is calculated, the measured corresponding to ranking anomalies point candidate area higher in the abnormal point ranking step based on the multi-path data viewing including the step of performing the image reconstruction processing has the data compression step, A step of extracting single path data by a plurality of transmission / reception antenna element pairs from detection data by a multipath system that receives a reflected wave by a combination of arbitrary transmission / reception antenna elements of the array antenna, and the extracted single path data And a step of compressing data by adding the signal strengths of the two . After the processing in the abnormal part ranking step is performed, a step of performing ranking display may be included.

  According to the present invention, an abnormal location candidate area is extracted by removing a normal location from the reflection intensity distribution with a threshold, and an evaluation value indicating the degree of abnormality is obtained from the abnormal location candidate region and the reflection intensity included therein. Then, the position of the abnormal part candidate area is ranked by the evaluation value, the measurement data necessary for the image reproduction process is identified, the image reproduction process is performed for the specific area, and the image reproduction process is performed. By narrowing down the area, the time from measurement to diagnosis can be greatly shortened, and the locations to be diagnosed are limited, so that the burden on the operator can be reduced.

Hereinafter, specific embodiments of a method and program for extracting an abnormal part candidate inside a structure according to the present invention will be described in detail with reference to the accompanying drawings.
FIG. 2 shows an example of the overall configuration of a ground penetrating radar system used in the embodiment. This system is an apparatus for three-dimensionally imaging defects inside a concrete floor slab using microwaves. As shown in FIG. 2, the array antenna 10, the high-frequency circuit 12, and the antenna elements are connected to each other. A power supply control logic circuit 14 and a switching circuit 16, a radar apparatus main body 20 having a measurement unit 18 including a measurement computer using a monitor, and an offline analysis processing unit 22 are configured.

  The radar system is characterized in that n transmitting antenna elements 24 and 26 receiving antenna elements 26 are arranged in a line as shown in FIGS. 3 (1) and 3 (2). An arbitrary one and an arbitrary combination of n receiving antenna elements can be selected. FIG. 3A is an example in which the transmission antenna elements 24 and the reception antenna elements 26 are alternately arranged in a line one by one. FIG. 3B is an example in which the transmission antenna elements 24 are arranged in a line and the reception antenna elements 26. In this example, the lines arranged in a line are shifted in the front-back direction (traverse direction). In order to perform the process of selecting all the combinations as the transmitting / receiving antenna element pairs, the switching circuit 16 sends a signal to the switching means described later, sends a microwave from the selected transmitting antenna element 24, The receiving circuit is switched so that all the receiving antenna elements 26 receive the reflected wave during transmission of the reflected wave from the selected transmitting antenna 24, and the receiving operation by all the receiving antenna elements 26 is completed. Later, the transmission antenna 24n (n = 1, 2, 3...) Is switched to the adjacent transmission antenna 24n + 1. These move together to cover the upper plane of the region to be measured by moving in the traverse direction 25 indicated by the arrow orthogonal to the array antenna 10 of FIG.

Data (reflected signal intensity) measured while traversing with the array antenna in the radar system described above can be expressed as φ (x, y ₁ , y ₂ , z). Where φ: reflected signal intensity, x: traverse coordinates, y ₁ : transmitting antenna element coordinates, y ₂ : receiving antenna element coordinates, z: depth coordinates.

  For this reason, when a wide range such as a tunnel is measured, the number of data points becomes enormous, and the obtained measurement data also becomes enormous. In order to efficiently obtain abnormal part candidates from this enormous amount of measurement data, it is necessary to process a large amount of measurement data in a short time.

  For this reason, in this embodiment, processing as shown in FIG. 1 is performed. A computer program for performing this processing is configured as follows. First, multipath data is measured using the radar system described above (step 100). When the measurement data is accumulated, an operation for extracting candidates for abnormal points from the data is performed (step 200). As described in the subroutine of FIG. 1, in this abnormal part extraction operation, first, the measurement data is compressed while the features remain, and the amount of data to be processed is reduced (step 210). .

なお、上式の記号の意味は以下の通りである。 The data compression processing with this characteristic left is to extract single-pass measurement data from the obtained multi-path measurement data, and to perform addition processing within a predetermined block range with the signal intensity as it is. That is, a signal obtained by the first transmission / reception antenna elements, a signal obtained by the second transmission / reception antenna elements, a third, and so on, are measured by a combination of adjacent transmission / reception antenna elements. Data (reflection signal intensity) is taken out. This can be easily performed because the single path data is simply extracted from the measured multipath data. Note that the pairs of transmission / reception antenna elements forming the single path data may not have the same number order as long as the pair relationship is constant. Then, the reflected signal intensity of the obtained single-pass measurement data is ψ (x, y, z), and the compressed data is ψ (X, Y, Z). Compressed signal.

The meanings of the symbols in the above formula are as follows.

  x: sensor traverse direction in pre-compression data, y: sensor train direction in pre-compression data, z: depth direction in pre-compression data, X: sensor traverse direction in post-compression data, Y: sensor train direction in post-compression data , Z: depth direction in post-compression data, l: size in x direction of range (block) to which pre-compression data is added, m: size in y direction of range (block) to which pre-compression data is added, n: pre-compression The size in the z direction of the range (block) to which data is added, o: the x direction shift amount of the range (block) to which the pre-compression data is added, p: the y direction shift amount of the range (block) to which the pre-compression data is added, q: z-direction shift amount of a range (block) to which pre-compression data is added.

FIG. 4 is an explanatory diagram of data compression. In order to facilitate the explanation, it is shown in two dimensions (X, Y). In this example, l = 4, m = 2, o = 2, and p = 1. In this example, the value of ψ (1,1) is ψ (1,1), ψ (2,1), ψ (3,1), ψ (4,1), ψ (1,2), ψ From (2,2), ψ (3,2), and ψ (4,2), the value of ψ (2,1) is ψ (3,1), ψ (4,1), ψ (5,1 ), Ψ (6,1), ψ (3,2), ψ (4,2), ψ (5,2), ψ (6,2). Explaining with the illustrated symbols, the compressed data a ₀ is an addition value of absolute values of the uncompressed data A _{0 to 3} and B _{0 to 3} , and the compressed data a ₁ is two data in the traverse direction. This is an added value of the absolute values of the shifted data A _{2 to 5} and B _{2 to 5} before compression. Here, the original data of Ψ (1,1) and Ψ (2,1) are overlapped even if there is a defect on the boundary of the original data area when not overlapping, the evaluation value shown later In order to reduce the area due to the division of the abnormal part candidate area at the time of calculation, the evaluation value is lowered even though it is originally one area, and the abnormal part candidate area is not properly ranked. .

  Next, an abnormal part candidate region is specified based on the distribution of the reflected signal intensity included in the compressed measurement data (step 220). The reflected signal intensity included in the obtained abnormal location candidate area (data reflection is not performed but the reflected signal intensity itself is a value equivalent to this) is totaled and used as an evaluation value for each area (step 230).

The abnormal part candidate area extraction process and the evaluation value calculation process here are performed as follows.
The reflected signal intensity has a distribution as shown in FIG. Since the abnormal part has a higher reflection intensity than the normal part, it is highly likely that the abnormal part exists in the right end portion of the distribution. Therefore, the normal location and the abnormal location are separated according to the threshold value from the reflected signal intensity distribution. The threshold is determined empirically, for example, 80% of the maximum signal strength is determined as the threshold.

ただし、Ｅｖａ：評価値 Ｖ：異常箇所候補領域である。 According to the threshold value, the compressed measurement data is binarized into a normal part and an abnormal part, and the obtained abnormal part candidate area is labeled to distinguish the abnormal part detection candidate area, and the reflected signal intensity of the labeled area is determined. Aggregated from the measurement data compressed based on the following mathematical formula 2, and set as an evaluation value of the abnormal part candidate.

However, Eva: evaluation value V: abnormal part candidate area.

  The defects inside the concrete are roughly classified into defects that spread in the depth direction such as junkers and those that are not large in the depth direction but have a large spread in the plane. For the former (defects extending in the depth direction), the reflected signal intensity of the radar is strong, but for the latter, the reflected signal intensity of the radar is not strong, so it is easy to miss.

  However, even if it is thin in the depth direction, a defect having a large spread in the plane cannot be overlooked. For this reason, the total reflection intensity is used as an evaluation value not only for areas where the reflection intensity is strong, but also for those that have weak reflection signals over a wide range (when the area average value is used, the reflection intensity Strong places will be ranked higher). However, it is not limited to this depending on the purpose of detecting the abnormal part. Furthermore, since the threshold value is determined based on the maximum value of the reflected signal distribution shown in FIG. 5, when there is a particularly strong reflected signal, the case where the weak reflected signal is widely spread may be below the threshold value and missed. Therefore, the evaluation value of the abnormal part candidate is calculated by dividing the area of the same area and limiting the influence of the maximum reflected signal intensity, which is the basis of the threshold, within the area. As a delimiter, for example, the measurement data is divided into areas per unit length in the longitudinal direction. In order to find a defect of about 10 cm, the unit length is preferably about 1 to 2 m, but can be about 10 m for the purpose of reducing the burden on the operator.

  Then, the abnormal part candidate areas are ranked according to the evaluation value (step 240), and the ranked abnormal part candidate list is obtained. The higher the candidate, the higher the possibility of including an abnormal part.

  Next, the abnormal location candidate list is displayed together with the tunnel surface image and the like (step 250), and the location of the abnormal location candidate is presented and the image reproduction process is assisted. For example, if you click on the abnormal location ranking shown in the table format, a mark will appear at the corresponding location on the tunnel table screen image, and image reproduction processing will start for the data in the range including that position. Linked with work. Further, if there is already obtained information (such as a sound hitting test result), it may be displayed simultaneously.

  When extraction of abnormal part candidates is performed in this way, it is cut out from multipath measurement data at a higher place in the abnormal part candidate list (step 300), and image reproduction processing is performed based on the cut out data (step 400). The operator diagnoses the presence or absence of an abnormal part from the obtained transmission image (step 500). If there is no more data in the abnormal part candidate list, the process ends (step 800). If there is any data in the abnormal part candidate list, the process returns to the next data extraction process (step 300). (Step 400) is repeated. If no abnormal location is found at a certain place, image reproduction processing and abnormal location diagnosis are not performed for the subsequent data (step 700).

  In this way, by omitting image reproduction processing and abnormality diagnosis of areas that do not include abnormal parts, it is possible to reduce the measurement time required to obtain a transmission image for all measurement data and to diagnose the transmission image by the operator. Realize the reduction. Moreover, the time from measurement to diagnosis result output can be shortened.

6 to 10 show data when measured with a multipath antenna.
The measurement data (FIG. 6) obtained by the array antenna is compressed (FIG. 7) while retaining the characteristics. From the compressed data distribution (FIG. 8), data that may contain an abnormal part is extracted, and an abnormal part candidate region (FIG. 9) is obtained. The reflection signal intensities included in the respective areas are aggregated to obtain an evaluation value, which is ranked according to the evaluation value and displayed as a list together with the compressed measurement data, the tunnel surface image, and the like (FIG. 10).

  From the top, measurement data in a range of 2.5 m (maximum length that can be analyzed by the computer used for image reproduction processing) including the abnormal part candidate region is cut out and image reproduction processing is performed to obtain a three-dimensional transmission image.

  When there is only data measured with a single path antenna, the following may be performed. Data measured along lattice points by a single path method such as a ground penetrating radar is processed in the same manner as in the case of multipath, and compressed as shown in FIG. After that, it is the same as in the case of multipath. This is because, in the operation of compressing multipath data, the result of taking out the measurement data by the combination of adjacent transceivers is equal to the result of measuring along the lattice points by the single path method.

  As described above, in this embodiment, the measurement data is compressed, the abnormal part candidate area is specified, and the reflected signal intensity included in the obtained abnormal part candidate area is totalized and evaluated, and the abnormal part candidate areas are ranked. Since the image reproduction process is performed from the measurement data of the upper place, a transmission image for all measurement data is obtained by omitting the image reproduction process and abnormality diagnosis of the area not including the abnormal part. Reduction of the measurement time until the operator, and the labor of diagnosing the transmitted image by the operator. Moreover, the time from measurement to diagnosis result output can be shortened.

  It can be used for non-destructive inspection of structures.

It is a flowchart which shows the abnormal location candidate extraction method inside the structure of embodiment. It is a block diagram of the underground radar system used for the embodiment. It is an arrangement form figure of the array antenna of the radar system. It is explanatory drawing of data compression. Conceptual diagram of reflection intensity distribution It is a depth cross section with the three-dimensional data measured by the array antenna. It is a depth cross section with the three-dimensional data which compressed the measurement data of FIG. 8 is a histogram of reflected signal intensity included in FIG. This is the result of leaving only the region including the abnormal part in FIG. FIG. 9 shows the result of calculating the evaluation value according to the area of FIG. 9 and the ranking display based on the measurement data (compression) of FIG. It is a flowchart which shows the conventional abnormal location diagnostic method using a three-dimensional imaging radar apparatus.

Explanation of symbols

10 ......... Array antenna, 12 ......... High frequency circuit (microwave circuit), 14 ......... Control logic circuit (system control circuit), 16 ......... Switching circuit (antenna switch), 18 ......... Measurement , 20... Radar device body, 22... Analysis processing unit (signal / image processor), 24... Transmitting antenna element, 26.

Claims (6)

A method of detecting an abnormal location by performing an internal inspection of a structure with a ground penetrating radar,
Compression of measurement data,
Identification of abnormal part candidate area,
Aggregating and evaluating the reflected signal intensity included in the obtained abnormal location candidate area,
Rank the abnormal candidate area,
There line image reproduction process from the measurement data of the location of the upper,
The measurement data is compressed by using an array antenna in which a plurality of transmission / reception antenna elements are arranged, detecting data by a multipath system that receives a reflected wave with an arbitrary combination of transmission / reception antenna elements, and then detecting a plurality of data from the detected multipath data. A method for detecting an abnormal location inside a structure, wherein single-path data from a pair of transmitting and receiving antenna elements is extracted and data compression is performed by adding signal strength . The abnormal part is identified by extracting a normal part candidate area by removing a normal part from the reflection intensity distribution with a threshold value, and using the abnormal part candidate area and the reflection intensity included therein, an evaluation value indicating the degree of abnormality is obtained. 2. The method for detecting an abnormal location inside a structure according to claim 1, wherein the location of the abnormal location candidate area is ranked by the evaluation value and specified. The image reproduction process from the measurement data, according to claim 1 or 2, characterized in that the upper ranking anomaly location candidate region based on the single path data based on the multi-path data in order Abnormal location detection method inside the structure. The abnormal location inside the structure according to any one of claims 1 to 3 , wherein if no abnormal location is found at a certain rank, image reproduction processing and abnormal location diagnosis are not performed for subsequent data. Detection method. It is a program to detect the abnormal part by inspecting the inside of the structure with a ground penetrating radar
Measuring multipath data from an array antenna; and
And data compression step of compressing the multipath data that the measurement,
Extracting an abnormal part candidate region from the compressed data;
Calculating an evaluation value aggregates reflected signal intensity included in the extracted abnormal point candidate area,
An abnormal part ranking step for ranking the abnormal part candidate regions based on the calculated evaluation value ;
Look including the step of performing the image reconstruction processing based on the multi-path data the measured corresponding to ranking Anomalies point candidate area higher in the abnormal place ranking step,
The data compression step includes a step of extracting single path data by a plurality of transmission / reception antenna element pairs from detection data by a multipath scheme that receives a reflected wave by a combination of arbitrary transmission / reception antenna elements of the array antenna; And a step of compressing the data by adding the signal strength of the extracted single path data . 6. The program for detecting an abnormal location inside a structure according to claim 5 , further comprising a step of displaying a ranking after the processing in the abnormal location ranking step is performed.

Images