JP2014010091A - Refuse state detection device in refuse incinerator - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a refuse state detection device capable of accurately detecting a refuse state such as a refuse level in a container for storing refuse in a refuse incinerator.SOLUTION: The refuse state detection device includes: a photographic image acquisition section 31 for acquiring a photographic image obtained by photographing the inside of a container 12 for storing refuse for supplying refuse to a furnace body of a refuse incinerator by a camera 27 for photographing the inside of the container at the interval of 1 second for instance; and a refuse level determination section 32 for determining a refuse level as the surface height of refuse by an FCM discriminator 33 using a fuzzy c-means method from image data acquired by the photographic image acquisition section 31.

Description

  The present invention relates to a waste state detection device for detecting a state of waste such as a waste level in a waste storage container in a waste incinerator.

  Conventionally, waste incinerators have been provided with a waste storage container for storing the waste to be put into the furnace, and the surface of the waste in the waste storage container in order to stabilize combustion in the furnace. The garbage level, which is the height, is kept constant. For this reason, the garbage level in the container for garbage storage was detected.

  Conventionally, when detecting the dust level, an ultrasonic level meter has been used, and the distance to the dust surface has been measured using ultrasonic waves (for example, see Patent Document 1).

JP 2008-64361 A

  However, when an ultrasonic level meter is used, the reflectivity of the ultrasonic wave differs depending on the type of dust thrown in and the surface shape of the dust, leading to measurement errors and allowing only partial detection of dust. First, if dust or the like is unevenly deposited on the detection portion, it leads to erroneous detection. Therefore, it is necessary for the operator to check the captured image of an industrial camera or the like each time.

  Such an operation has led to a delay in the introduction of garbage and to the occurrence of garbage withering in the combustion chamber. An industrial camera may be used instead of the ultrasonic level meter. However, in order to detect the garbage level, edge processing is performed on the captured image to obtain the boundary between the wall of the garbage storage container and the garbage. Although it is conceivable, there is a case where a long thing remains in the waste storage container by being caught in the vertical direction, and in this case, it may not be detected by edge processing, and the waste is thrown in. The garbage level may not be detected accurately due to the image of the crane or the shade of light or shadow.

  SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a waste state detection device capable of accurately detecting a waste state such as a waste level in a waste storage container in a waste incinerator.

In order to solve the above problems, a waste state detection apparatus for a waste incinerator according to claim 1 of the present invention takes an image of the inside of a waste storage container for supplying waste into the furnace body of the waste incinerator with a photographing camera. A captured image acquisition unit that acquires the captured image at predetermined time intervals;
The apparatus includes a dust level determining unit that determines a dust level that is the surface height of the dust from the image data acquired by the photographed image acquiring unit using a classifier using a fuzzy c-average method.

Moreover, the waste state detection apparatus in the refuse incinerator according to claim 2 of the present invention is the waste state detection apparatus according to claim 1,
The waste level obtained by the waste level judgment unit is input in chronological order, and a waste level difference calculation unit for obtaining a waste level difference that is a difference between adjacent waste level values, and a waste level difference calculation unit A bridge detection means comprising a bridge determination unit for determining the presence or absence of a bridge based on the garbage level difference is provided.

Furthermore, the waste state detection apparatus in the refuse incinerator according to claim 3 of the present invention is the waste state detection apparatus according to claim 1 or 2,
A waste level average value calculator that calculates the average value of a predetermined number of waste levels by inputting the waste levels obtained by the waste level judgment unit in time series, and a burnout inspection that detects the burnout point in the furnace body The apparatus is provided with a garbage withering predicting means comprising an output part and a garbage withering judgment part for inputting the obtained average value of the garbage level and the burn-out point and predicting withering of the garbage.

  According to the configuration of the dust state detection apparatus, the classifier based on the fuzzy c average method is used to detect the dust level by the dust level detection means. The level can be detected accurately.

  In addition, the level of dust in the waste storage container is detected by a discriminator using the fuzzy c-average method, and the presence / absence of a bridge is judged based on this dust level, so that the occurrence of a bridge can be known more accurately. Can do.

  Furthermore, since the garbage withering is predicted based on the dust level detected by the classifier using the fuzzy c-average method, the garbage withering can be predicted more accurately.

It is a mimetic diagram showing a schematic structure of a refuse incinerator concerning an embodiment of the invention. It is a block diagram which shows schematic structure of the waste condition detection apparatus in the waste incinerator which concerns on the same Example. It is a block diagram which shows the structure of the garbage state detection apparatus. It is a block diagram which shows schematic structure of the FCM discrimination device used with the garbage state detection apparatus. It is explanatory drawing of the fuzzy clustering by the FCM discriminator. It is explanatory drawing of the classification by the FCM discriminator. It is a graph which shows the comparison of the value of the garbage level using the same FCM discriminator, and the measurement result using an ultrasonic sensor.

Hereinafter, a dust state detection apparatus in a waste incinerator according to an embodiment of the present invention will be described with reference to FIGS.
First, a schematic configuration of a waste incinerator provided with the waste state detection device according to the present invention will be described.

  As shown in FIG. 1, the waste incinerator is, for example, a stoker furnace, and in the furnace main body 1, a waste supply port 2 is located at the front, a grate 3 at the center, and an incineration residue outlet 4 at the rear. Combustion chambers 5 each provided to combust dust and provided with a freeboard space above the front side are provided, and a first flue 6 for guiding the exhaust gas generated in the combustion chamber 5 to the outside and A second flue 8 in which the heat recovery boiler unit 7 is arranged is provided.

The waste incinerator includes a waste supply device 11 for supplying waste into the furnace body 1, a control device (not shown) that controls combustion control, a waste input command, and the like.
The waste supply device 11 includes an inverted square pyramid (hopper shape) waste storage container 12 provided with an opening 12a in the lower part and a waste storage space part 12b on the upper side, and the waste storage container 12. A waste pushing device 13 for pushing the waste in the space portion 12b to the furnace body 1 side, and a bridge for the waste generated in the side wall of the waste storage container 12 and generated in the container. It is comprised from the bridge | braking cancellation | release apparatus 14 for eliminating.

  The bridge release device 14 includes an oscillating plate 15 provided in an opening 12c provided in a side wall portion so as to be oscillatable in a vertical plane (indicated by an arrow a), and the oscillating plate 15 for storing garbage. The cylinder device 16 is configured to swing in the inner and outer directions of the container 12. Therefore, if the cylinder device 16 is operated to swing the swing plate 15 into or out of the garbage storage container 12, that is, the bridge is eliminated.

Further, the waste incinerator is provided with a waste state detection device according to the present invention.
As shown in FIG. 2, the waste state detection device 21 includes a waste level detection means (also referred to as a detection device) 22 for detecting a waste level that is the height of the waste surface in the waste storage container 12, and the waste. A bridge detection means (also referred to as a detection device) 23 for detecting whether or not a bridge is generated in the waste storage container 12 based on the dust level detected by the level detection means 22, and the dust level detection means 22. And a garbage withering predicting means (also referred to as a prediction device) 24 for predicting the occurrence of garbage withering based on the dust level detected by the above and the burnout point (burnout position) on the grate 3 in the combustion chamber 5. Yes.

  Further, as shown in FIG. 1, in order to detect the dust state, that is, the combustion state of the dust and the dust level in the dust storage container 12, the rear wall portion 1 a of the combustion chamber 5 has a furnace in which the front is photographed. An imaging camera (for example, an industrial camera such as a CCD camera is used) 26 is provided, and an in-container imaging camera (for example, an industrial camera such as a CCD camera) that images the interior of the waste storage container 12 is used. ) 27 is provided. The interior of the combustion chamber 5 is photographed from the rear wall side (incineration residue outlet side) to the front wall side (garbage supply port side) by the in-furnace photographing camera 26, and by the in-container photographing camera 27, The interior of the waste storage container 12 is photographed from the rear wall side (furnace main body side) downward to the front wall side.

  As shown in FIG. 3, the dust level detection means 22 is a camera that acquires the above-described in-container photographing camera 27 and a photographed image, that is, a still image photographed by the in-container photographing camera 27 at, for example, one second intervals. The image acquisition unit 31 includes a dust level determination unit 32 that inputs image data acquired by the photographed image acquisition unit 31 and determines a dust level that is a dust surface height based on a fuzzy c average method. .

  The bridge detection means 23 inputs the dust level obtained by the dust level determination unit 32 in a time series and detects whether or not a bridge has occurred (whether or not a bridge has occurred). A garbage level difference calculation unit 41 for obtaining a difference between levels (garbage level inclination), and a dust level difference obtained by the garbage level difference calculation unit 41 are input to generate a bridge. It is composed of a bridge determination unit 42 for determining an abnormality such as a wasteful fall.

  The bridge determination unit 42 determines whether the dust level difference falls within a plurality of ranges, for example, three ranges. For example, two threshold values (also referred to as setting values) t1 and t2 are obtained in advance as a range of a dust level difference that can be determined that no bridge has occurred by experiment or the like, and a dust level difference is in a range of 0 to t1. If it is within, it is determined that a bridge has occurred and its index value is set to “1”. If the dust level difference is within the range of t1 to t2, it is determined that no bridge has occurred and the index value is set to “0.5”. Further, when the garbage level difference exceeds t2, it is determined that the garbage has fallen and its index value is set to “0”. Note that a membership function can also be used to determine whether or not a bridge is generated from the dust level difference.

  Then, when finally determining (determining) the presence or absence of the occurrence of bridges, for example, determination is made for 10 garbage level data obtained in time series. That is, ten garbage level difference index values are held in the memory, and the ten index values are multiplied. For example, if the value S of the multiplied index value is equal to or greater than a predetermined value (for example, 0.8), it is determined that a bridge has occurred, and the value S of the multiplied index value is within a predetermined range ( For example, if 0 <S <0.8), it is determined to be normal, and if the index value S multiplied by 0 is 0 (zero), an abnormality such as sudden drop has occurred. It is judged. Thus, by multiplying the index values for a plurality of data, it is determined whether or not the same level is continuously maintained.

  In the garbage withering predicting means 24, the dust level obtained by the dust level judging unit 32 is inputted in time series, and the occurrence of the garbage withering in the combustion chamber 5 is generated based on the burnout point on the grate 3. is expected.

  The garbage withering prediction unit 24 acquires (inputs) a still image in the combustion chamber 5 captured by the in-furnace imaging camera 26 at predetermined time intervals, for example, at intervals of 1 second, and a combustion chamber image acquisition unit 51. From the image data acquired by the combustion chamber image acquisition unit 51, image processing such as edge processing is performed to detect the burnout point detection unit 52, and the dust level obtained by the dust level determination unit 32 is determined as time. A waste level average value calculation unit 53 for obtaining an average value of dust levels in a predetermined time by inputting in series, and a burnout point obtained by the waste level average value calculation unit 53 and the burnout point detection unit 52 are input. And the prediction judgment part 54 which estimates generation | occurrence | production of garbage withering is comprised.

  In the prediction determination unit 54, when the average value of the garbage level is lower than a predetermined threshold value (set value) and the burnout point is closer to the front of the grate 3 (closer to the garbage supply port 2), A prediction that “occurs” is output.

  By the way, the dust level detection means 22 performs fuzzy clustering using a fuzzy c-means (also referred to as FCM discrimination method). 33, which is also configured by software).

Here, general classification by the FCM discriminator 33 will be described.
Here, in order to simplify the description, the description will be made on the assumption that the classes are divided into two classes and each class is also divided into two clusters.

When this fuzzy c-average method is applied, there are a preliminary process for selecting various parameters and an actual process for actual identification.
In the preliminary process, the training data is divided into two classes, and each class is also divided into two clusters.

  Then, the parameters of the membership function (described later) are optimized using the cluster center position (hereinafter referred to as cluster center), the data such as the variance-covariance matrix and the evaluation data obtained at this time. . The training data is image data obtained by photographing the surface of the garbage at an arbitrary height in the garbage storage container 12, and the evaluation data is also image data obtained by photographing the interior of the garbage storage container 12 with a photographing camera. It is.

Next, the fuzzy c averaging method will be described.
Since this fuzzy c-average method uses an iteratively weighted least square method, the objective function J _i is set as shown in the following equation (1).

但し、（１）式中、ｕ_ｋｉはメンバーシップ、Ｄはマハラノビス距離で下記（２）式にて表わされる。

In the equation (1), _uki is a membership, and D is a Mahalanobis distance, which is represented by the following equation (2).

Ｓ_ｉは分散共分散行列（ファジィ分散共分散行列）、ｖ_ｉはデータの平均値つまりクラスター（ｉ）の中心で、それぞれ下記（３）式および（４）式にて表わされる。なお、ｋはデータ番号である。

S _i is a variance-covariance matrix (fuzzy variance-covariance matrix), and v _i is an average value of data, that is, the center of cluster (i), and is represented by the following formulas (3) and (4), respectively. Note that k is a data number.

また、クラスター（ｉ）の混合比率αを下記（５）式にて表わす。

Further, the mixing ratio α of the cluster (i) is expressed by the following formula (5).

上記（２）式〜（５）式を（１）式で示す目的関数値が収束するまで繰り返すのがＦＣＭ識別器によるクラスタリングである。

Clustering by the FCM discriminator repeats the above equations (2) to (5) until the objective function value indicated by equation (1) converges.

  In the preliminary process, the cluster phase for calculating the Mahalanobis distance and the first phase of clustering the training data for each class to obtain the variance-covariance matrix, and the optimization of the membership function parameters are performed. Therefore, a second phase for classifying the evaluation data for each class is provided.

By the way, the membership (value) (tilde u _qk ) to the class q of the k-th data x _k is expressed by the following equation (6). In the formula, a tilde symbol (˜) is added to the head of u, and this “tilde u” is expressed as “* u” in the text. The membership function is expressed as “u ^* ” as shown below.

（６）式中、ｕ_ｑｋｊ ^＊はクラスターｊに対するメンバーシップ関数で、下記（７）式で表わされる。

In the equation (6), u _qkj ^* is a membership function for the cluster j and is represented by the following equation (7).

π _q is a mixing ratio of class q (mixing ratio of training data), that is, a prior probability, c is the number of clusters j for each class, and is 2 (c = 2) in this embodiment.

上記（７）式のメンバーシップ関数ｕ_ｑｋｊ ^＊には複数のパラメータが含まれており、これらのパラメータは、訓練用データおよび評価用データを用いて（訓練用データだけを用いてもよい）最適化が行われる。

The membership function u _qkj ^* in the above equation (7) includes a plurality of parameters, and these parameters are optimal using training data and evaluation data (only training data may be used). Is done.

  Then, when classifying, semi-hardware clustering is performed in order to shorten the calculation time. In this semi-hardware clustering, discrete values (for example, 0.6, 0.8, etc.) are used as membership.

  In this semi-hardware clustering, classes are divided into two and each class is divided into two clusters. This is because experience has shown that dividing each class into two clusters allows efficient clustering. The number of clusters is not limited to two, and may be one or three or more, for example. For example, when the discrete value is 0.5, the number of clusters is 1, and when the discrete value is 1.0, two hard (not fuzzy) clusters are obtained. Although the number of the classes has been described as two (although it has been described), of course, it may be three or more, and in the case of this embodiment described in detail later, the number is five.

In the first phase, semi-hardware clustering is performed for each class using training data.
That is, in order to perform semi-hard clustering between hardware and fuzzy, membership u _ki is set as in the following equation (8).

Hereinafter, a specific procedure will be described (the number of clusters (i) is two). Note that β is specified by the discrete value described above.
First, the initial membership (u _k1 , u _k2 ) for each cluster is set as in the following equations (9) and (10).

なお、上記式中、ｆ_ｋは１つのクラスの画像データｘ_ｋの主成分得点である。

In the above formula, f _k is a main component score of one class of image data x _k .

  When membership is given by the above equation, v is obtained from equation (4), S is obtained from equation (3), and Mahalanobis distance D is obtained from equations (2) and (3), (8) Update membership u with formula. This update is repeated until membership has converged.

Next, clustering is performed for each class (classes 1 and 2) using the training data. That is, each class is divided into two clusters.
At this time, semi-hardware clustering is performed using the objective function J.

That is, the membership u for each training data that minimizes the objective function J is obtained. Thus, the center v of each cluster and its membership u are obtained.
Then, based on the obtained membership u, clustering is performed for each class.

In the second phase, clustering is performed using the evaluation data.
That is, the Mahalanobis distance D is obtained for the center v of each cluster obtained in the first phase. At this time, the membership u obtained in the first phase is used.

Then, by applying (using) the obtained Mahalanobis distance D to the membership function u ^* shown by the predetermined equation (6), each parameter (m, γ, ν) in the membership function u ^* is used. , Α; these are also called free parameters or hyperparameters). In this optimization, a particle swarm optimization method (PSO) is used.

Here, this particle swarm optimization method will be briefly described.
In this particle swarm optimization method, the cluster mixing ratio α, which is the same parameter as the three parameters (m, γ, ν) in the membership function u ^* , is optimized.

  That is, in this particle group optimization method, the search for the best position of the particle group is performed by the following update formulas (11) and (12).

上記式中、Paraは粒子の位置を示すパラメータで、α，ｍ，γ，νからなるベクトルである。Veloは粒子の速度ベクトルである。Randは「０」と「１」との間の乱数の対角行列である。ｗ_０，ｃ_１，ｃ_２はスカラー定数である。pbset とgbest はそれぞれpbset とgbestの位置ベクトルである。

In the above formula, Para is a parameter indicating the position of the particle and is a vector composed of α, m, γ, and ν. Velo is the velocity vector of the particle. Rand is a diagonal matrix of random numbers between “0” and “1”. w ₀ , c ₁ , and c ₂ are scalar constants. pbset and gbest are position vectors of pbset and gbest, respectively.

These free parameters are determined so as to minimize the misidentification rate of the evaluation data.
That is, optimization is performed so that the membership * u using the membership function u ^* belongs to the correct class, in other words, the correct class membership * u is larger.

In the preparation step described above, the membership function u ^* is obtained.
Next, an actual process for detecting a dust level using the FCM discriminator will be briefly described.

  Here, a description will be given assuming that the garbage level is divided into five ranges. For example, as shown in FIG. 6, the garbage level in the garbage storage container 12 is 0 to 1 (class 1: center value is 0.5), 1-2 (class 2: center value is 1.5). 2 to 3 (class 3: center value is 2.5), 3 to 4 (class 4: center value is 3.5), and 4 to 5 (class 5: center value is 4.5) To do. In addition, the garbage level of 0-5 is shown in the part of the container 12 for garbage storage of FIG.

First, the Mahalanobis distance D with respect to the center v of each cluster obtained in the preparation step of image data that has been photographed and subjected to predetermined processing (described later) is obtained, and this Mahalanobis distance is expressed by the above equation (7). The membership function value u ^* obtained by the substitution is substituted into the above equation (6), and the membership * u for each cluster in each class is obtained. Then, the membership * u obtained for each cluster is added to obtain the membership * u _i for each class (i).

When the membership * u _i is determined in each class, the sum of values membership value is applied to the central value of each class (median) is obtained. Of course, the total sum calculation unit (executed by software) is provided for determining the total sum of values membership * u _i is applied to the central value.

More specifically, based on FIG. 6, for example, membership * u _{1 of} class 1 (center value is 0.5) is 0 (zero) and class 2 (center value is 1) for the image data to be detected. .5) membership * u ₂ is 0.01, class 3 (central value is 2.5) membership * u ₃ is 0.66, class 4 (central value is 3.5) membership * u _{When 4} is 0.033 and the membership * u _{1 of} class 5 (center value is 4.5) is 0 (zero), a numerical value represented by the following formula is obtained as the membership.

0.5 × 0 + 1.5 × 0.01 + 2.5 × 0.66 + 3.5 × 0.33 + 4.5 × 0 = 2.82
That is, 2.82 is obtained as the output value. If this value is converted into an actual garbage height, the garbage level can be expressed by a fine value. Of course, this output value may be used as it is.

Here, the configuration of the FCM classifier described above will be described.
As shown in FIG. 4, the FCM discriminator includes a normalization unit 61 that partitions the image data acquired by the captured image acquisition unit 31 into, for example, 10000 (100 × 100 pixels), and the normalization unit 61 normalizes the data. A vectorizing unit 62 that converts the image data into a 10,000-dimensional vector, a dimension compressing unit 63 that compresses the dimension of the vector data vectorized by the vectorizing unit 62 into, for example, 50-dimensional data by principal component analysis, A Mahalanobis distance calculation unit 64 for obtaining a Mahalanobis distance with respect to the center of a cluster obtained in advance by the dimension compression unit 63 and a membership function u based on the Mahalanobis distance obtained by the Mahalanobis distance calculation unit 64. ^* and membership * Request u (7) and (6) member which expression is provided -Ship calculation unit 65, and a level determination unit 66 for determining which class representing the garbage level the image data belongs to based on the membership * u obtained by the membership calculation unit 65. . In the calculation by the dimension compression unit 63, the Mahalanobis distance calculation unit 64, the membership calculation unit 65, and the level determination unit 66, an appropriate shape / size data and image of the waste storage container 12 from the database unit (not shown). The compression coefficient, identification parameters, etc. are read and used. In actual operation, the identification parameter read is a value that is optimally adjusted by training or the like (the description is omitted, but a parameter adjustment unit is provided).

A specific calculation procedure will be described. For example, when there are two classes to be divided, the Mahalanobis distance calculation unit 64 calculates the Mahalanobis distance D with respect to the cluster center obtained in advance of the image data, and the membership calculation unit 65 A membership function u ^* (function value) is obtained based on the Mahalanobis distance D, and a membership * u is obtained based on the membership function u ^* . That is, membership * u ₁ , * u ₂ is obtained for each cluster. Then, for each class, membership is added, and memberships * u _c1 and * u _c2 in the class are obtained.

Then, the level determination unit 66 compares the obtained memberships * u _c1 and * u _{c2 in} each class, and determines that this image data belongs to the larger one.

For example, the membership in each class at the time of this determination is shown in FIG. 5, the two Class 1 and Class 2, in which respectively form two clusters 1 and cluster 2, respectively, the class 1 side, there is a cluster 2 clusters 1 and v ₁₂ of v _11, also On the class 2 side, there is also a cluster 1 of v ₂₁ and a cluster 2 of v ₂₂ . In FIG. 5, assuming that the point X is image data to be determined, the membership of X in one cluster 1 in class 1 is u ₁₁ , the membership of X in the other cluster 2 is u ₁₂ , and the class 2, the membership of X to class 1 is u ₂₁ , and the membership of X to the other cluster 2 is u ₂₂ , the membership u ₁ to class 1 of X is u ₁₁ + u ₁₂ , membership _{u 2} for the class _2, the _u 21 ₊ u _22. That is, the total value of membership of each cluster in each class (which is the sum of membership heights for both clusters) becomes the membership for each class. Of the memberships u ₁ and u ₂ , they belong to the larger class.

  Note that the free parameters are searched so that the misidentification rate is minimized by the particle swarm optimization method described above. For example, when the cluster mixing ratio α is changed, the cluster region (shown by contour lines) in FIG. Will change, and therefore the boundaries of the two classes will also change non-linearly.

As described above, in the case of this embodiment having five classes, the classification is as shown in FIG.
Next, the operation of the waste incinerator using the waste state detection device 21 will be described.

  In the state in which various identification parameters are determined in advance by the training data and the evaluation data by the garbage level detection means 22, the interior of the garbage storage container 12 is captured by the in-container photographing camera 27, for example, at intervals of 1 second. Thus, a still image is acquired.

Then, the still picture is inputted to the dust level determining unit 32, where along with the classification is performed in FCM discriminator 33, based on membership * u _i obtained at that time, required dust level It is done.

FIG. 7 shows the result of comparing the dust level (shown by a solid line) by the FCM discriminator 33 and the measurement value (shown by a broken line) by the ultrasonic sensor.
As shown by an arrow E in FIG. 7, the measurement value obtained by the ultrasonic sensor sometimes outputs an erroneous measurement value, but the dust level detected by the FCM discriminator 33 is stable, and the actual value It almost coincided with the garbage level.

That is, the dust level detection means 22 using the FCM discriminator 33 can measure the dust level more accurately than the conventional ultrasonic level meter.
On the other hand, the dust level is detected every predetermined time, and data of the dust level is input to the bridge detection means 23 in time series.

  In the dust level difference calculation unit 41 of the bridge detection means 23, a difference between the data before and after the dust level continuously input in time series is obtained, and this dust level difference is input to the bridge determination unit 42. It is compared with a preset threshold value to determine whether a bridge has occurred, a normal state in which no bridge has occurred, or an abnormal state in which some drop has occurred.

If it is determined that a bridge has occurred, a release command is output from the bridge detection means 23 to the bridge release device 14 of the waste supply device 11 to cancel the bridge.
Further, the image captured by the in-furnace imaging camera 26 is input to the combustion chamber image acquisition unit 51 and then input to the burnout point detection unit 52, where image processing such as edge processing is performed, A burnout point on the grate 3 is detected. For the burn-out point, a value obtained by other detection means, for example, a burn-out point detected by an FCM discriminator may be used.

On the other hand, the garbage levels obtained in time series are input to the garbage level average value calculation unit 53, and an average value for a predetermined number (for example, 10) is obtained.
Then, the dust level average value and the detected burnout point are input to the prediction determination unit 54, and it is predicted whether or not the garbage wither will occur.

  For example, when the dust level is low (for example, when it is in the range of “0 to 1” shown in FIG. 1) and the burnout point is near the front of the grate 3 (when close to the dust supply port 2 side) It is predicted that garbage will die.

  In this case, for example, a garbage input command into the garbage storage container 12 is output to the garbage input crane via the central controller of the waste incinerator. Of course, when a bridge is generated, the bridge release device 14 is instructed to perform a bridge release operation via the central controller.

  In this way, the dust level in the waste storage container 12 is detected by the FCM discriminator 33, and whether or not a bridge has occurred is determined based on this dust level. You can know the occurrence. Similarly, since the FCM discriminator 33 predicts the garbage withering based on the dust level, the garbage withering can be predicted more accurately.

The configuration of the FCM discriminator described above will be generally and simply described as follows.
This FCM discriminator vectorizes a normalization unit that obtains and normalizes a plurality of representative values from image data taken by a photographing camera, and image data composed of the representative values obtained by the normalization unit. A vectorization unit, a dimension compression unit that compresses the dimension of the vector data vectorized by the vectorization unit, and a Mahalanobis distance to the cluster center obtained from the training data of the vector data compressed by the dimension compression unit in advance Substituting the Mahalanobis distance calculation unit for obtaining the value and the Mahalanobis distance obtained by the Mahalanobis distance calculation unit into the membership function (u ^* ) shown in the following equation (13) and substituting this function value into the following equation (14): The membership calculation unit for obtaining membership (tilde u) and the membership calculated by this membership calculation unit A photographed image with a flop is obtained by including a example five classes level determining section divided into (class showing the different ranges of dust level) (also say state determining unit).

但し、上記式中、α_ｑｊはクラスｑ内のクラスターｊの混合比率、Ｓは分散共分散行列、π_ｑはクラスｑの混合比率である。

In the above formula, α _qj is the mixing ratio of cluster j in class q, S is the variance-covariance matrix, and π _q is the mixing ratio of class q.

Also, the principal component analysis method is used when the dimension is compressed by the dimension compression unit.
Furthermore, a particle swarm optimization method is used when optimizing each parameter (m, γ, ν, α) in the membership function (u ^* ) expressed by equation (13).

DESCRIPTION OF SYMBOLS 1 Furnace body 2 Garbage supply port 3 Grate 4 Incineration residue outlet 5 Combustion chamber 11 Garbage supply device 12 Garbage storage container 12a Opening part 12b Garbage storage space 13 Garbage extrusion device 14 Bridge release device 21 Garbage state detection device 22 Waste level detection means 23 Bridge detection means 24 Waste withering prediction means 26 In-furnace imaging camera 27 In-container imaging camera 31 Captured image acquisition section 32 Waste level determination section 33 FCM discriminator 41 Waste level difference calculation section 42 Bridge determination section 51 Combustion chamber image acquisition unit 52 Burnout point detection unit 53 Waste level average value calculation unit 54 Prediction determination unit 61 Normalization unit 62 Vectorization unit 63 Dimension compression unit 64 Mahalanobis distance calculation unit 65 Membership calculation unit 66 Level determination unit

Claims (3)

A captured image acquisition unit that acquires, with a predetermined time interval, captured images obtained by capturing the interior of a waste storage container for supplying waste into the furnace body of the waste incinerator,
A waste incinerator comprising: a waste level judging unit for judging a waste level which is the surface height of the waste from an image data obtained by the photographed image obtaining unit by a discriminator using a fuzzy c-average method. Waste state detection device.   The waste level obtained by the waste level judgment unit is input in chronological order, and a waste level difference calculation unit for obtaining a waste level difference that is a difference between adjacent waste level values, and a waste level difference calculation unit 2. A garbage state detection apparatus in a refuse incinerator according to claim 1, further comprising bridge detection means comprising a bridge judgment unit for judging whether or not there is a bridge based on a garbage level difference.   A waste level average value calculator that calculates the average value of a predetermined number of waste levels by inputting the waste levels obtained by the waste level judgment unit in time series, and a burnout inspection that detects the burnout point in the furnace body The waste withering predicting means comprising an output part and a withering judgment part for predicting the withering of garbage by inputting the average value of the found garbage level and the burn-out point is provided. The garbage state detection apparatus in the refuse incinerator described.

Images