JP2007055541A - Vehicular ride quality diagnosing device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a vehicular ride quality diagnosing device capable of correctly diagnosing the vehicular ride quality according to the human feeling. <P>SOLUTION: The vehicular ride quality diagnosing device comprises an acceleration detector for detecting the acceleration waveform of a vehicle, an A/D converter 3 for outputting the time series data Ei by sampling the acceleration waveform, a wavelet conversion means 5 for outputting a plurality of resolutions K0 to Kj-1 as a set of the result of the continuous wavelet conversion for each basis function by performing the continuous wavelet conversion to the time series data by a plurality of basis functions obtained by enlarging or contracting the predetermined mother wavelet function to the predetermined times, an index calculation means for calculating the predetermined number of indexes to evaluate the resolutions, an index storage means for storing the reference index value calculated from the normal acceleration waveform, a comparator for comparing the index calculated by the index calculation means with the reference index value, and a determination device 7 for outputting the sate of the acceleration waveform based on the output of the comparator. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a vehicle riding comfort diagnostic apparatus that evaluates the riding comfort of a vehicle using an acceleration waveform of the vehicle during traveling.

  In the conventional evaluation of the riding comfort of a railway vehicle, an acceleration meter is installed in the railway vehicle to be diagnosed, and the effective value of the vertical acceleration of the railway vehicle at a specific time is measured. Then, the acceleration at this specific time is set as a judgment condition for the riding comfort of the iron vehicle, and the riding comfort of the railway vehicle is evaluated by comparing the effective value of the measured acceleration with a predetermined value by a computer or the like. (For example, refer nonpatent literature 1).

“Riding comfort evaluation of railway vehicles”, [online], Railway Technical Research Institute, Vehicle Structural Technology Research Department, Vehicle Movement Laboratory, [Search April 28, 2005], Internet <URL: http: // www . rtri. or. jp / rd / openpublic / rd41 / dynamics / research09_confort. html>

  In the conventional ride comfort evaluation of a railway vehicle, acceleration information other than the specific time is outside the determination condition as described above. For this reason, even if there is no problem in acceleration at a specific time, the ride comfort may be deteriorated, and there is a problem that the ride comfort of the railway vehicle cannot be accurately evaluated according to human sensitivity.

  An object of the present invention is to solve the above-described problems, and an object of the present invention is to provide a vehicle ride comfort diagnosis capable of accurately diagnosing the ride comfort of a vehicle in accordance with human sensitivity. To provide an apparatus.

  A vehicle riding comfort diagnostic apparatus according to the present invention includes an acceleration detection means for detecting an acceleration waveform of a vehicle, a sampling means for sampling the acceleration waveform every predetermined time and outputting time series data, and a predetermined mother wavelet function. Wavelet transform operation means for performing continuous wavelet transform on time series data by a plurality of basis functions expanded or reduced by a factor of two and outputting a plurality of resolutions that are a set of continuous wavelet transform results for each basis function; Index calculation means for calculating a predetermined number of indices for evaluating resolution, index storage means for storing a reference index value calculated by the index calculation means from a normal acceleration waveform, and an index calculated by the index calculation means and index storage means The comparison means for comparing the reference index value stored in the memory, and the state of the acceleration waveform based on the output of the comparison means It is obtained by a judging means for outputting.

  According to the vehicle riding comfort diagnostic apparatus of the present invention, the acceleration waveform is obtained from the resolution K0 to Kj-1 of the continuous wavelet transform using the French Hat function of the acceleration waveform based on the time series data output from the sampling means as the mother wavelet function. The average value that can be used as an index to determine whether it is normal is calculated and patterned, and compared with the average value when the acceleration waveform is normal. Riding comfort can be accurately diagnosed according to human sensitivity.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the drawings, the same or corresponding members and parts will be described with the same reference numerals.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing a vehicle riding comfort diagnostic apparatus according to Embodiment 1 of the present invention. FIG. 2 is a block diagram showing the diagnostic device 4 of FIG.

  In FIG. 1, this vehicle ride comfort diagnostic apparatus is provided in a railway vehicle, detects an acceleration of the vehicle and outputs an acceleration waveform (not shown), and an acceleration waveform detected. A filter 1 that passes a predetermined frequency band component that is arbitrarily set, an amplifier 2 that amplifies and outputs an acceleration waveform that has passed through the filter 1 with a predetermined amplification factor that is arbitrarily set, and is output from the amplifier 2 The A / D converter 3 (sampling the acceleration waveform at an arbitrarily set predetermined time, performing analog-digital conversion, and outputting time-series data Ei (i = 0, 1,..., L−1) ( Sampling means) and a diagnostic device 4 that determines the state of the acceleration waveform based on the time-series data Ei and outputs an alarm to the outside as necessary.

  In FIG. 1, the acceleration waveform is shown as one, but actually, it is assumed that vertical, horizontal, and longitudinal acceleration waveforms are included.

  In FIG. 2, this diagnostic device 4 includes a wavelet transform computing unit 5 (wavelet transform computing unit), a basis function computing unit 6, average value computing units Q0 to Qj-1 (index calculating unit), and setting units R0 to R0. Rj-1 (index storage means), comparators S0 to Sj-1 (comparison means), pattern determiner 7 (determination means), pattern setter 8, and alarm generator 9 (alarm output means) Have.

  The diagnostic device 4 is configured by a microprocessor (not shown) having a storage unit storing a program and a CPU. Each block constituting the diagnostic device 4 is stored as software in the storage unit.

The wavelet transform computing unit 5 computes a continuous wavelet transform for a plurality (j) of basis functions obtained by enlarging or reducing the mother wavelet function to an arbitrary predetermined multiple using the French Hat function as a mother wavelet function, and for each basis function. A set of wavelet coefficients at the same sampling time as the input signal obtained in (1) is obtained. Here, K represents one of j basis functions 0, 1,..., J−1. A set of wavelet coefficients at the same sampling time as the input signal Ei of 0, 1,..., L−1 by the continuous wavelet transform on the basis function K is referred to as resolution K. The wavelet transform computing unit 5 outputs resolution K0, resolution K1,..., Resolution Kj-1.
The basis function calculator 6 gives a basis function to the wavelet transform calculator 5.

The average value calculators Q0 to Qj-1 respectively calculate the average values (indexes) of the resolutions K0 to Kj-1 output from the wavelet transform calculator 5.
The setters R0 to Rj−1 include, for example, reference average values calculated from respective resolutions obtained by continuous wavelet transform using a French Hat function based on a normal acceleration waveform as a reference obtained from a new vehicle as a mother wavelet function. Reference index values) are stored in advance.
Comparators S0 to Sj-1 compare the average value output from average value calculators Q0 to Qj-1 with the reference average value stored in setting devices R0 to Rj-1.

  In FIG. 2, three resolutions K are representatively shown as resolution K0, resolution K1, and resolution Kj-1, but in reality, j resolutions K0 to Kj-1 are obtained. Similarly, three average value calculators Q are representatively shown as an average value calculator Q0, an average value calculator Q1, and an average value calculator Qj−1. An average value calculator Q is required. Similarly, the setters R0 to Rj-1 and the comparators S0 to Sj-1 require j setters R and comparators S, respectively.

The pattern determination unit 7 compares the allowable pattern set by the pattern setting unit 8 in which the allowable pattern in the normal acceleration waveform serving as a reference is stored with the output of the comparators S0 to Sj−1 to determine the acceleration waveform. Output the status.
The alarm generator 9 outputs an alarm to the outside as needed based on the output of the pattern determiner 7.

The operation of the vehicle riding comfort diagnostic apparatus having the above configuration will be described below.
First, the acceleration waveform output from the acceleration detector is input to the A / D converter 3 via the filter 1 and the amplifier 2.
The acceleration waveform is sampled at predetermined time intervals by the A / D converter 3 and converted from analog to digital, and output as time-series data Ei (i = 0, 1,..., L−1).
The time series data Ei is data sampled over the entire operation time.

  When the time series data Ei is input to the diagnosing device 4, the time series data Ei becomes the input signal of the wavelet transform computing unit 5 shown in the following equation (1) as it is.

    Input signal X (t), t = 0, 1,..., L−1 (1)

  The French Hat function, which is a mother wavelet function used by the wavelet transform computing unit 5, is a function expressed by the following equation (2).

  The 0th basis function W (t) is defined as the following equation (3) using D which is a predetermined value larger than 3.

    W (t) = φ ((2 * D / L) * t−D) (3)

  Here, the basic formula of the continuous wavelet transform will be described. When the input function X (t), the scale factor a, the shift factor b on the time axis, the mother wavelet function W (t), and the continuous wavelet transform result Y (a, b), the relationship of the following equation (4) is established. .

Since the input signal X (t) is sampled time series data, the shift coefficient b is obtained with respect to the sampling time point. The scale factor should be obtained continuously for all values, but it takes too much time. Therefore, the scale coefficient is obtained only with j representative values.
Here, the basis function Pn (t) is obtained by the following equation (5), assuming that j scale factors are a geometric sequence of the common ratio α (real number other than 1).

Pn (t) = W (t * α ⁿ ) (5)

In equation (4), assuming that the shift coefficient b is adjusted so that φ = 0 when t = 0, Y (a, b) when a = α ⁿ is expressed as Yn (t). When the integral is replaced with the sum, the following equation (6) is obtained.
Note that [X] is a Gaussian sign that means the maximum integer that does not exceed the real number X.

However, t = 0, 1,..., L−1, and n = 0, 1,.
Since X (t) has a value only discretely, t ′ is replaced with t ″ as in the following equation (7) so that it can be referred to when it has a value. Also, t <0 and t For> L, X (t) = 0.

t ′ = t ″ + (L / α ⁿ ) / 2 − [(L / α ⁿ ) / 2] (7)

Here, if β = (L / α ⁿ ) / 2 − [(L / α ⁿ ) / 2], the equation (6) can be expressed as the following equation (8).

Here, a set of Yn (t) calculated by Expression (8) at time t = 0, 1,..., L−1 is the resolution Kn. D and α are adjusted so that the riding comfort judgment is appropriately performed.
Note that the resolution may be obtained by linearly interpolating the value of X (t) using Expression (6), or the resolution may be obtained by setting β = 0 in Expression (8). Even in these cases, the error is small in many cases.

The resolutions K0 to Kj−1, which are the calculation results of the continuous wavelet transform using the French Hat function obtained as described above as the mother wavelet function, are converted into the following equation (9) by the average value calculators Q0 to Qj−1. As shown, each average value μ ₁ ⁰ , μ ₁ ¹ ,..., Μ ₁ ^j−1 is calculated.

Subsequently, when the average values μ ₁ ⁰ , μ ₁ ¹ ,..., Μ ₁ ^j−1 are calculated by the average value calculators Q0 to Qj−1, the above averages are calculated by the comparators S0 to Sj−1. The values μ ₁ ⁰ , μ ₁ ¹ ,..., Μ ₁ ^j−1 and the reference average values stored in the setting devices R0 to Rj−1 are compared.

  Here, the setters R0 to Rj-1 store average values of resolutions K0 to Kj-1 of continuous wavelet transform using the French Hat function of a normal acceleration waveform as a reference as a mother wavelet function. You may make it memorize | store the normal value at the time of normal which the external apparatus (not shown) calculated.

Next, the above-mentioned average values μ ₁ ⁰ , μ ₁ ¹ ,..., Μ ₁ ^j−1 and the reference average values stored in the setting devices R ⁰ to Rj−1 are obtained by the comparators S ^{0 to} Sj−1. For example, when the above average values μ ₁ ⁰ , μ ₁ ¹ ,..., Μ ₁ ^j−1 exceed three times the reference average value stored in the setting devices R0 to Rj−1, the comparison result Is determined to be abnormal, and a signal indicating that it is abnormal is output to the pattern determiner 7.

  Here, the resolution K0 to Kj-1 of the continuous wavelet transform using the French Hat function of the acceleration waveform as the mother wavelet function is a filter for each corresponding resolution. In addition, since the average value is an index indicating the average value of the probability density distribution function, the average value of the filtered time series data is regularly distributed around a certain value when the acceleration waveform is normal. . Here, when the acceleration waveform is not normal, the average value of the specific resolution K or all of the resolutions K0 to Kj-1 deviates from the average value when it is normal, and thus shows a distribution different from the standard distribution. Since this deviation depends on an acceleration waveform that is not normal, whether or not the acceleration waveform is normal can be determined by comparing the average values.

  Next, the pattern determination unit 7 compares the determination results of the comparators S0 to Sj-1 with the allowable pattern stored in advance in the pattern setting unit 8, and determines and outputs the final acceleration waveform state. . This determination condition is a condition that, for example, if any one of the resolutions K0 to Kj-1 is abnormal, the final determination is abnormal, and is stored in the pattern setting unit 8 in advance according to the target.

  Subsequently, when a signal indicating that the final acceleration waveform state is abnormal is input to the alarm generator 9 from the pattern determiner 7, the alarm generator 9 informs the outside that an abnormal acceleration waveform has been generated. In order to clarify, a display device (not shown) indicates that an abnormality has occurred, a monitoring device (not shown) or the like outputs a signal indicating that an abnormality has occurred, and the series of processing is completed. The

According to the vehicle riding comfort diagnostic apparatus according to Embodiment 1 of the present invention, the resolution of continuous wavelet transform using the French Hat function of the acceleration waveform based on the time-series data output from the A / D converter 3 as the mother wavelet function. Since the average value that can be an index for determining whether or not the acceleration waveform is normal is calculated and patterned from K0 to Kj-1, and compared with the average value when the acceleration waveform is normal, the specific value is specified. It is possible to accurately diagnose not only the time but also the ride comfort of the entire operation time according to the sensitivity of the person, and the situation can be determined at a higher speed than the conventional one.
In addition, the calculation time can be further shortened by adopting the average value.

Embodiment 2. FIG.
FIG. 3 is a block diagram showing a diagnostic device 4A of the vehicle riding comfort diagnostic apparatus according to Embodiment 2 of the present invention.
Here, for the same type as in the first embodiment, “A” is appended after the same reference numeral, and detailed description thereof is omitted.
Note that the vehicle riding comfort diagnostic apparatus according to the present embodiment is provided with a diagnostic device 4A shown in FIG. 3 instead of the diagnostic device 4 of FIG.

In FIG. 3, in this diagnostic device 4A, instead of the average value calculators Q0 to Qj-1 in the first embodiment, the variances (indexes) of the resolutions K0 to Kj-1 output from the wavelet transform calculator 5 are used. Distributed computing units T0 to Tj-1 (index calculation means) are provided.
Further, in the setting devices RA0 to RAj-1, there are reference variances (reference index values) calculated from the respective resolutions obtained by continuous wavelet transform using a French Hat function based on a normal acceleration waveform as a reference as a mother wavelet function. Each is stored in advance.
Comparators SA0 to SAj-1 compare the dispersion output from dispersion calculators T0 to Tj-1 with the reference dispersion stored in setting devices RA0 to RAj-1.
Other configurations are the same as those in the first embodiment, and the description thereof is omitted.

The operation of the vehicle riding comfort diagnostic apparatus having the above configuration will be described below.
Note that the description of the same operation as in the first embodiment is omitted.
The resolutions K0 to Kj−1 obtained by the wavelet transform computing unit 5 by the computation of the continuous wavelet transform using the French Hat function as the mother wavelet function are represented by the following formula (10) by the dispersion computing units T0 to Tj−1. Thus, the variances μ ₂ ⁰ , μ ₂ ¹ ,..., Μ ₂ ^j−1 are calculated.

Subsequently, when the dispersion μ ₂ ⁰ , μ ₂ ¹ ,..., Μ ₂ ^j−1 are calculated by the dispersion calculators T0 to Tj−1, the above-mentioned dispersion μ ₂ is calculated by the comparators SA0 to SAj−1. ⁰ , μ ₂ ¹ ,..., Μ ₂ ^j−1 and the reference dispersion stored in the setting devices RA0 to RAj−1 are compared.

  Here, the setters RA0 to RAj-1 store variances of resolutions K0 to Kj-1 of continuous wavelet transform using the French Hat function of a normal acceleration waveform as a reference as a mother wavelet function. The normal variance calculated by the device (not shown) may be stored.

Next, the above-mentioned dispersions μ ₂ ⁰ , μ ₂ ¹ ,..., Μ ₂ ^j−1 and the reference dispersion stored in the setting devices RA0 to RAj−1 are compared by the comparators SA0 to SAj−1. For example, when the above-mentioned dispersion μ ₂ ⁰ , μ ₂ ¹ ,..., Μ ₂ ^j−1 exceeds three times the reference dispersion stored in the setting devices RA0 to RAj−1, the comparison result is abnormal. And a signal indicating the abnormality is output to the pattern determiner 7.

  Here, the resolution K0 to Kj-1 of the continuous wavelet transform using the French Hat function of the acceleration waveform as the mother wavelet function is a filter for each corresponding resolution. Further, since the variance is an index indicating the variance of the probability density distribution function, the variance of the filtered time series data is regularly distributed around a certain value when the acceleration waveform is normal. Here, when the acceleration waveform is not normal, the distribution of the specific resolution K or all of the resolutions K0 to Kj-1 deviates from the distribution when the acceleration is normal, and thus shows a distribution different from the standard distribution. Since this deviation depends on the acceleration waveform that is not normal, whether or not the acceleration waveform is normal can be determined by comparing the variances.

According to the vehicle riding comfort diagnostic apparatus according to Embodiment 2 of the present invention, the resolution of the continuous wavelet transform using the French Hat function of the acceleration waveform based on the time-series data output from the A / D converter 3 as the mother wavelet function. Since the variance that can be an index for determining whether or not the acceleration waveform is normal is calculated and patterned from K0 to Kj-1, and compared with the variance when the acceleration waveform is normal, the configuration is compared with a specific time. In addition, the ride comfort of the vehicle during the entire operation time can be accurately diagnosed according to the sensitivity of the person, and the situation can be determined at a higher speed than the conventional one.
Further, by adopting dispersion, it is possible to improve the sensitivity of a phenomenon in which an abnormality appears in the variation.

Embodiment 3 FIG.
FIG. 4 is a block diagram showing a diagnostic device 4B of the vehicle riding comfort diagnostic apparatus according to Embodiment 3 of the present invention.
Here, for the same type as in the first embodiment, “B” is appended after the same reference numeral, and detailed description is omitted.
The vehicle riding comfort diagnostic apparatus according to the present embodiment is provided with a diagnostic device 4B shown in FIG. 4 instead of the diagnostic device 4 of FIG.

In FIG. 4, this diagnosing device 4B includes a skewness (index) of resolutions K0 to Kj-1 output from the wavelet transform computing unit 5 instead of the average value computing units Q0 to Qj-1 in the first embodiment. Distortion calculators U0 to Uj-1 (index calculation means) are provided.
The setters RB0 to RBj-1 include a reference skewness (reference index value) calculated from each resolution obtained by continuous wavelet transform using a French Hat function based on a normal acceleration waveform serving as a reference as a mother wavelet function. Are stored in advance.
Comparators SB0 to SBj-1 compare the skewness output from skewness calculators U0 to Uj-1 with the reference skewness stored in setting devices RB0 to RBj-1.
Other configurations are the same as those in the first embodiment, and the description thereof is omitted.

The operation of the vehicle riding comfort diagnostic apparatus having the above configuration will be described below.
Note that the description of the same operation as in the first embodiment is omitted.
The resolutions K0 to Kj−1 obtained by the wavelet transform computing unit 5 by the computation of the continuous wavelet transform using the French Hat function as the mother wavelet function are expressed by the following equation (11) by the skewness computing units U0 to Uj−1. As shown, each skewness μ ₃ ⁰ , μ ₃ ¹ ,..., Μ ₃ ^j−1 is calculated.

Subsequently, skewness _mu ^{3 0} by skewness computing unit ^{_{U0~Uj-1, μ 3 1,}} ···, μ 3 when ^j-1 is calculated, the comparator SB0~SBj-1, the above strain The degrees μ ₃ ⁰ , μ ₃ ¹ ,..., Μ ₃ ^j−1 are compared with the reference skewness stored in the setting devices RB0 to RBj−1.

  Here, the setters RB0 to RBj-1 store the skewness of the resolution K0 to Kj-1 of continuous wavelet transform using the French Hat function of the normal acceleration waveform as a reference as the mother wavelet function. You may make it memorize | store the normal skewness computed by the external apparatus (not shown).

Next, by the comparators SB0 to SBj-1, the above-mentioned skewness μ ₃ ⁰ , μ ₃ ¹ ,..., Μ ₃ ^j-1 and the reference skewness stored in the setting devices RB0 to RBj-1 are obtained. If, for example, the above-mentioned skewness μ ₃ ⁰ , μ ₃ ¹ ,..., Μ ₃ ^j−1 exceeds three times the reference skewness stored in the setting devices RB0 to RBj−1, the comparison result Is determined to be abnormal, and a signal indicating that it is abnormal is output to the pattern determiner 7.

  Here, the resolution K0 to Kj-1 of the continuous wavelet transform using the French Hat function of the acceleration waveform as the mother wavelet function is a filter for each corresponding resolution. In addition, since the skewness is an index indicating the skewness of the probability density distribution function, the skewness of the filtered time series data is regularly distributed around a certain value when the acceleration waveform is normal. . Here, when the acceleration waveform is not normal, the skewness of the specific resolution K or all of the resolutions K0 to Kj-1 deviates from the skewness when the acceleration is normal, and thus the distribution is different from the standard distribution. This deviation depends on the acceleration waveform that is not normal. Therefore, it is possible to determine whether or not the acceleration waveform is normal by comparing the skewness.

According to the vehicle riding comfort diagnostic apparatus according to Embodiment 3 of the present invention, the resolution of the continuous wavelet transform using the French Hat function of the acceleration waveform based on the time series data output from the A / D converter 3 as the mother wavelet function. Since the degree of distortion that can be used as an index for determining whether or not the acceleration waveform is normal is calculated from K0 to Kj−1 and patterned, and compared with the degree of distortion when the acceleration waveform is normal. It is possible to accurately diagnose not only the time but also the ride comfort of the entire operation time according to the sensitivity of the person, and the situation can be determined at a higher speed than the conventional one.
In addition, by adopting the skewness, it is possible to improve the sensitivity of a phenomenon in which an abnormality appears in the bias.

Embodiment 4 FIG.
FIG. 5 is a block diagram showing a diagnostic device 4C of the vehicle riding comfort diagnostic apparatus according to Embodiment 4 of the present invention.
Here, about the same kind as Embodiment 1, "C" is attached after the same numerals, and detailed explanation is omitted.
The vehicle riding comfort diagnostic apparatus according to the present embodiment is provided with a diagnostic device 4C shown in FIG. 5 instead of the diagnostic device 4 of FIG.

In FIG. 5, this diagnosing device 4C includes a kurtosis (index) of resolutions K0 to Kj-1 output from the wavelet transform computing unit 5 instead of the average value computing units Q0 to Qj-1 in the first embodiment. Kurtosis calculators V0 to Vj-1 (index calculation means) are provided.
The setters RC0 to RCj-1 include a reference kurtosis (reference index value) calculated from each resolution obtained by continuous wavelet transform using a French Hat function based on a normal acceleration waveform as a mother wavelet function. Are stored in advance.
Comparators SC0 to SCj-1 compare the kurtosis output from kurtosis calculators V0 to Vj-1 with the reference kurtosis stored in setting devices RC0 to RCj-1.
Other configurations are the same as those in the first embodiment, and the description thereof is omitted.

The operation of the vehicle riding comfort diagnostic apparatus having the above configuration will be described below.
Note that the description of the same operation as in the first embodiment is omitted.
The resolutions K0 to Kj−1 obtained by the wavelet transform calculator 5 by the calculation of the continuous wavelet transform using the French Hat function as the mother wavelet function are expressed by the following equation (12) by the kurtosis calculators V0 to Vj−1. As shown, each kurtosis μ ₄ ⁰ , μ ₄ ¹ ,..., Μ ₄ ^j−1 is calculated.

Subsequently, when the kurtosis values μ ₄ ⁰ , μ ₄ ¹ ,..., Μ ₄ ^j−1 are calculated by the kurtosis calculators V0 to Vj−1, the above-described kurtosis is calculated by the comparators SC0 to SCj−1. The degrees μ ₄ ⁰ , μ ₄ ¹ ,..., Μ ₄ ^j−1 and the reference kurtosis stored in the setting devices RC0 to RCj−1 are compared.

  Here, the setters RC0 to RCj-1 store the kurtosis of the resolution K0 to Kj-1 of the continuous wavelet transform using the French Hat function of the normal acceleration waveform as a reference as the mother wavelet function. You may make it memorize | store the normal kurtosis computed by the external apparatus (not shown).

Next, the kurtosis μ ₄ ⁰ , μ ₄ ¹ ,..., Μ ₄ ^j−1 and the reference kurtosis stored in the setting devices RC0 to RCj−1 are obtained by the comparators SC0 to SCj−1. For example, if the kurtosis μ ₄ ⁰ , μ ₄ ¹ ,..., Μ ₄ ^j−1 described above exceeds three times the reference kurtosis stored in the setting devices RC0 to RCj−1, the comparison result Is determined to be abnormal, and a signal indicating that it is abnormal is output to the pattern determiner 7.

  Here, the resolution K0 to Kj-1 of the continuous wavelet transform using the French Hat function of the acceleration waveform as the mother wavelet function is a filter for each corresponding resolution. In addition, since kurtosis is an index indicating the kurtosis of the probability density distribution function, the kurtosis of time-series data that has been filtered is regularly distributed around a certain value when the acceleration waveform is normal. . Here, when the acceleration waveform is not normal, the kurtosis of the specific resolution K or all of the resolutions K0 to Kj-1 deviates from the kurtosis in the case of normal, and thus shows a distribution different from the standard distribution. Since this deviation depends on an acceleration waveform that is not normal, whether or not the acceleration waveform is normal can be determined by comparing the kurtosis.

According to the vehicle riding comfort diagnostic apparatus according to Embodiment 4 of the present invention, the resolution of continuous wavelet transform using the French Hat function of the acceleration waveform based on the time-series data output from the A / D converter 3 as the mother wavelet function. Since the kurtosis that can be used as an index for determining whether or not the acceleration waveform is normal is calculated and patterned from K0 to Kj-1, and compared with the kurtosis when the acceleration waveform is normal, It is possible to accurately diagnose not only the time but also the ride comfort of the entire operation time according to the sensitivity of the person, and the situation can be determined at a higher speed than the conventional one.
In addition, by adopting kurtosis, it is possible to improve the sensitivity of a phenomenon in which an abnormality appears in the spread.

In the first to fourth embodiments, the French Hat function is used as the mother wavelet function. However, any other function can be used as long as it is a function that can be used for continuous wavelet transformation, such as a Mexican Hat function or a Gabor function. Also good.
In addition, the predetermined number (j) of scale coefficients for obtaining the wavelet transform results have a relation of the geometric series, but may not necessarily have the relation of the geometric series. The scale factor may be determined in any way as long as a predetermined number of wavelet transform results in different frequency regions are obtained.

  In the first to fourth embodiments, only one of the average value, variance, skewness, and kurtosis is used. However, a plurality of these may be used to determine whether the acceleration waveform is normal. Further, whether or not the acceleration waveform is normal may be determined based on a weighted sum of at least two of average value, variance, skewness, and kurtosis.

1 is a block diagram illustrating a vehicle riding comfort diagnostic apparatus according to Embodiment 1 of the present invention. It is a block diagram which shows the diagnostic device of FIG. It is a block diagram which shows the diagnostic device of the vehicle riding comfort diagnostic apparatus which concerns on Embodiment 2 of this invention. It is a block diagram which shows the diagnostic device of the vehicle riding comfort diagnostic apparatus which concerns on Embodiment 3 of this invention. It is a block diagram which shows the diagnostic device of the vehicle riding comfort diagnostic apparatus which concerns on Embodiment 4 of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Filter, 2 Amplifier, 3 A / D converter (sampling means) 4, 4A-4C diagnostic device, 5 Wavelet transformation computing unit (wavelet transformation computing means), 6 Basis function computing unit, 7 Pattern judgment unit (determination means) ), 8 pattern setter, 9 alarm generator (alarm output means), K0 to Kj-1 resolution, Q0 to Qj-1 average value calculator (index calculation means), R0 to Rj-1, RA0 to RAj-1 , RB0 to RBj-1, RC0 to RCj-1 setter (index storage means), S0 to Sj-1, SA0 to SAj-1, SB0 to SBj-1, SC0 to SCj-1 comparators (comparison means), T0 to Tj-1 dispersion calculator (index calculation means), U0 to Uj-1 skewness calculator (index calculation means), V0 to Vj-1 kurtosis calculator (index calculation means).

Claims (6)

Acceleration detecting means for detecting an acceleration waveform of the vehicle;
Sampling means for sampling the acceleration waveform every predetermined time and outputting time-series data;
A plurality of basis functions obtained by enlarging or reducing a predetermined mother wavelet function by a predetermined factor to perform continuous wavelet transform on the time-series data to obtain a plurality of resolutions as a set of continuous wavelet transform results for each basis function. Wavelet transform computing means to output;
Index calculation means for calculating a predetermined number of indexes for evaluating the resolution;
Index storage means for storing a reference index value calculated by the index calculation means from a normal acceleration waveform;
A comparison means for comparing the index calculated by the index calculation means with a reference index value stored in the index storage means;
A vehicle riding comfort diagnosis apparatus comprising: a determination unit that outputs a state of the acceleration waveform based on an output of the comparison unit.   The vehicle ride comfort diagnosis device according to claim 1, wherein the index includes any one of an average value, a variance, a skewness, and a kurtosis.   The vehicle ride comfort diagnosis apparatus according to claim 1, wherein the index is a weighted sum of at least two of an average value, variance, skewness, and kurtosis.   The vehicle ride comfort diagnosis apparatus according to claim 1, wherein the mother wavelet function is a French Hat function.   2. The vehicle riding comfort diagnosis apparatus according to claim 1, further comprising alarm output means for outputting an alarm to the outside based on an output of the determination means.   The vehicle ride comfort diagnosis device according to claim 1, wherein the vehicle is a railroad vehicle.

Images