JPH0121913B2 - - Google Patents

Description

[Detailed Description of the Invention] Conventionally, a method for extracting a peak value from the detected power of a measuring device is to sample the detected power S at a constant cycle, and the measured value extracted at each cycle changes from an increasing trend to a decreasing trend. There is a method of determining the measured value at the time of conversion as the peak value. However, according to this method, as is clear from Figure 1, when the noise N ₁ is superimposed on the upward slope of the detected power, the detected power temporarily starts to decrease from the peak of the noise, so this peak is set to the peak value. A problem occurs where it is determined that As a countermeasure for this, a standard value that takes into account fluctuations due to measurement errors.
_L0 was set in advance, and when the measured value followed a decreasing trend after the peak value was detected, the difference _d0 between the peak value and subsequent measured values was extracted by comparing the above reference value _L0 . There is a method of determining whether the peak value is due to the influence of noise or whether it is the true peak value of the detection power. In this case, if the above difference is larger than the reference value _L0 , it is determined that the detected peak value is due to the influence of noise, and the detected power is again in an increasing trend and the true peak value appears, and the difference is the reference value L If it is smaller than _0, it is determined that the detected peak value is the true peak value of the detected force and that it follows a decreasing direction. However, even with the above method, if the noise N ₂ is superimposed on the peak of the detection force, there is only one switch from an increasing trend to a decreasing trend, so the peak value of the noise N ₂ is judged to be the peak value, which is inaccurate. values are not measured. In view of the above points, the present invention proposes a method capable of stably extracting a peak value even if there is noise at the peak of the detection force. First, the principle of the present invention will be explained with reference to FIG. In the figure, the detected power S is sampled every t period, and the detected power changes from an increasing trend to a decreasing trend while sequentially comparing the measured values at adjacent periods t ₁ , t ₂ , t ₃ ...tn. The measurement value at the time of conversion is determined to be the peak value, which has not changed from before. however,
According to the present invention, the reference value L is set in advance in anticipation of fluctuations in the period t due to measurement errors, and in the comparison process in the area of increasing tendency, the difference between the measured values in adjacent periods is determined in advance. The feature is that it is determined whether or not the measured value obtained in the latest cycle is affected by noise by comparing the value L with the reference value L. For example, if the difference (D ₃ − D ₂ ) between the measured values at adjacent cycles t ₂ and t ₃ is smaller than the reference value L, the measured value D ₃ obtained at the t ₃ cycle is assumed to be unaffected by noise. However, it is determined that the difference between the measured value D _l in the t _l period and the measured value D _l-1 returned in the t _l-1 period is greater than the reference value L and is influenced by noise, and the measured value is D _l will not be adopted. Then, the measured value D _{l +1 of the next cycle t l+1} in which noise was detected is compared with the measured value D _{l -1 of the previous cycle t l-1} , and the difference is greater than twice the reference value L. If it is small, the measured value D _l+1 is assumed to be unaffected by noise, and if it is large, it is assumed that it is continuously affected by noise, and the measured value D _{l+2 with period t l +2} is used. The difference from the measured value D _l-1 is compared with a value three times the reference value L. In this way, when noise is detected in a certain cycle, the measured value in the previous cycle is held, and in each subsequent cycle, the difference between the detected measured value and this held measured value is calculated for each cycle. It is determined whether or not the signal is continuously affected by noise by sequentially comparing it with the reference value L which is doubled, tripled, quadrupled, etc., respectively. Furthermore, when the detected force changes from an increasing trend to a decreasing trend in a period tm, the measured value Dm in a period tm changes to a continuing period tm ₊₁・tm ₊₂ ...
Measured value Dm ₊₁・Dm ₊₂ …larger measured value Dm
is retained. Then, after comparing several times up to the period tn, if Dm is larger in all cases, then
This is determined as the peak value. According to the above method, for example, noise N ₂ is superimposed on the peak of the detection force, and after adopting the measured value Dm _-1 at period tm _-1 , the measured value D′m affected by the noise at period tm
Even if D′m and Dm -1 are detected, the difference between D′m and Dm _-1 is the reference value L
No larger value will be used for this measurement. And each measurement value at period tm ₊₁・tm ₊₂ ...tn
Dm ₊₁・Dm ₊₂ ...If Dm _-1 is larger than Dn, this value is taken as the peak value, so it is possible to obtain a result that is almost the same as the normal detection power. FIG. 3 shows a device embodiment of the present invention in the case of a coin sorting device used in a vending machine, and FIG. 4 shows a flowchart of its control operation. In Fig. 3, 1 is a coin path in which the detected coin 2 rolls;
Reference numeral 3 denotes a coin sorting sensor, to which the output of an oscillator 4 which oscillates at a constant frequency of 200 KHz or more during normal operation is applied, and an alternating magnetic field is formed by a coil 5. Then, the coin 2 rolls through the coin passage 1 and coils 5.
As the inductance approaches the coil 5, the inductance decreases and the oscillation frequency increases, and then as the coil 5 moves away from the coil 5, the inductance increases and the oscillation frequency decreases. The coin sorting device judges the suitability and type of the coin 2 based on the maximum value of the frequency change at this time, and the oscillation output of the oscillator 4 is sampled at a predetermined period of 1 ms, and the frequency at each period is determined. It is counted by the counter 6. Therefore, the sampling signal SP is input to one side of the AND gate 7, and the sampling signal SP
When the counter 6 is reset at the rising edge of the oscillator 4, the output of the oscillator 4 is introduced into the counter 6 during the sampling period, the number of peaks of the oscillation output is counted, and the frequency is detected. 8 is a rewritable memory with storage areas M ₀ , M ₁ , M ₂ , M ₃ , and M ₄ , where M ₀ contains the detected frequency in the latest cycle, and M ₁ contains the noise detected in the previous cycle. The unaffected frequency, M ₂ contains the difference between the frequencies stored in M ₀ and M ₁ ,
_M3 stores the total frequency when noise is detected in continuous cycles, and _M4 stores the number of cycles within the decreasing range when the detected frequency decreases. Reference numeral 9 is a memory for storing data for determination, and _M5 is set with the aforementioned reference value L, and _M6 is set with peak value end determination data, respectively. The storage areas of the memories 8 and 9 are specified by the memory control circuit 10, and reading or writing is controlled.
11 is an arithmetic circuit that calculates the difference between the values stored in M ₀ and M ₁ , and adds 1 to the values stored in M ₃ and M ₄ . And 12 is a comparison circuit
Is the value stored in M ₂ positive? Is the value stored in M ₂ larger than the product of the value of M ₃ plus ₁ and the reference value? It is determined whether each value is larger than the peak value end determination data. 13 is a write circuit that writes zero to _M3 or _M4 and further writes the value of _M0 to _M1 . Further, 14 is a control circuit that outputs the sampling signal.
SP is output to the AND gate 7 to sample the output of the oscillator 4 and output a control signal to sequentially execute the control shown in FIG. Next, the operation will be explained based on the flowchart shown in FIG. For example, the measured value of the frequency related to period a
When Da is detected by the counter 6, the control circuit 14 first specifies M ₀ using the memory control circuit 10 and writes Da into M ₀ (Da→M ₀ ). Then, the control circuit 14 sequentially specifies M ₀ and M ₁ and also outputs a control signal.
CT ₁ is output, and the arithmetic circuit 11 calculates the difference between Da stored in _M0 and the measured value Da _-1 at the period (a- ₁ ) stored in M1, and then the control circuit 14
specifies M ₂ and writes this difference to M ₂ (M ₀ −M ₁
→ _M2 ). Then, the control circuit 14 specifies M ₂ and outputs a control signal CT ₂ , and the comparison circuit 12 determines whether the value of M ₂ is greater than zero (M ₂ >0). As a result of this judgment, if M ₂ is larger than zero, a judgment signal J ₁ is input to the control circuit 14, that is, Da>Da _-1 , and it is found that the measured value is on an increasing trend.
specifies M ₂ , M ₃ , and M ₅ sequentially and also outputs the control signal.
CT ₃ is output, and the comparison circuit 12 determines whether the value stored in M ₂ is greater than the product of the value of M ₃ added by 1 and the reference value L set in M ₅ (M ₂ ＞
L. (M ₃ +1)). As a result, if M ₂ is large, a judgment signal J ₂ is input to the control circuit 14, and since Da is a measured value affected by noise, the control circuit 14 specifies M ₃ and outputs a control signal CT ₄ , and the arithmetic circuit 11 is 1 to the value of M ₃
are added and the addition result is written to _M3 . i.e. _M3
It is stored that one noise detection period has occurred, and the flow is to take in the measured value in the next period (a+1) (Da ₊₁ → M ₀ ). Also (M ₂ > L・(M ₃ +1))
If it is determined that M ₂ is small, a determination signal J ₃ is input to the control circuit 14, and assuming that Da is an appropriate measurement value that does not include noise, the control circuit 14 sequentially specifies M ₀ and M ₁ and sends a control signal CT ₅ The write circuit 13 outputs
Read the value of M ₀ and write it to M ₁ (M ₀ → M ₁ ), and hold the measured value Da in M ₁ . Next, the control circuit 14
The write circuit 13 designates _M3 and outputs the control signal _CT6 , and resets _M3 ( _M3 →0). That is, since _M3 stores the total number of noise cycles when noise is detected in consecutive cycles, it is reset when no noise is detected. and control circuit 1
4 specifies M ₄ and outputs a control signal CT ₇ ,
The write circuit 14 resets M ₄ (M ₄ →0).
In other words, M ₄ shows a decreasing tendency in each cycle, and M ₂
In order to remember the number of consecutive periods where is smaller than zero,
If the measured value shows an increasing trend without being affected by noise, it is reset. Furthermore, if it is determined that (M ₂ > 0), and M ₂ is small, a determination signal J ₄ is input to the control circuit 14, and it is found that the measured value shows a decreasing tendency, so the control circuit 14
Specifying _M4 and outputting the control signal _CT8 , the arithmetic circuit 11 adds 1 to the value of _M4 , and the addition result is
Written to M ₄ (M ₄ +1→M ₄ ). Next, the control circuit 14 sequentially designates M ₄ and M ₆ and sends the control signal.
CT ₉ is output, and the comparator circuit 12 compares the value of M ₄ with the peak value end judgment data A of M ₆ (M ₄ =
A). As mentioned earlier, _M4 stores the total number of cycles when the measured value in each cycle is in a continuous decreasing trend, and this number of cycles reaches a predetermined number (i.e., peak value end judgment data A). Then, the judgment signal _J5 is input from the comparison circuit 12, and the control circuit 14 determines the measured value stored in _M1 as the peak value, and if the value of M4 is small, the control circuit 14 determines the measurement value stored in M1 as the peak value, and if the value of _M4 is small, the control circuit 14 determines the measurement value stored in M1 to be the peak value. The measurement value will be taken in. According to the above configuration, for example, starting from period a, when the measured value shows a decreasing tendency, the measured value Da _-1 acquired in the previous period (a-1) is held in M ₁ , and from then on a+
1・a+2...If all the measured values in the period are smaller than Da _-1 , M ₁ will continue to hold Da _-1 and the period a
Measured value at period (a+A) after A period has elapsed
If Da _+A is also smaller than Da _-1 , this Da _-1 is determined as the peak value. Also, if noise is detected in the measured value Da at period a, the measured value Da _-1 at period (a-1)
is kept as it is at M ₁ , Da is excluded from comparison, and the measured value Da ₊₁ in the next cycle (a+1) becomes Da _-1
compared to According to the present invention described in detail above, instead of determining the influence of noise only after the measured value shows a decreasing tendency as in the past, the detected measured value is used for each sampling period to determine the noise that was previously adopted. Since noise is detected by comparing it with measured values that do not include it, stable noise detection is performed. Moreover, in the comparison operation, a measurement value at a cycle in which noise is detected is not employed, so that an inappropriate measurement value is prevented from being set as a peak value. Furthermore, when noise is detected, the subsequent comparison operation is performed based on the product of the total number of cycles from the noise detection cycle and the reference value, so even if noise spans consecutive cycles, it can be detected. At the same time, the noise is 2 cycles/3
There is an advantage that control becomes easy because there is no need to prepare reference values for each period.

[Brief explanation of drawings]

Fig. 1 shows the principle of conventional peak value detection, Fig. 2 shows the principle of the present invention, Fig. 3 shows an embodiment applied to a coin sorting device, and Fig. 4 shows a flowchart of control operation. . 1... Coin passageway, 3 ... Coin sorting sensor, 6
... Counter, 8, 9 ... Memory, 11 ... Arithmetic circuit, 12 ... Comparison circuit, 13 ... Write circuit, 1
4...Control circuit.

Claims (1)

[Claims] 1. Sampling the detected force of the measuring device at regular intervals and sequentially comparing the measured values in adjacent cycles, determining the measured value when the detected force changes from an increasing trend to a decreasing trend as the peak value. Regarding the method, a reference value is set in anticipation of fluctuations in one cycle due to measurement errors, and in the comparison process in the area of increasing trend, the difference between measured values in adjacent cycles is smaller than the reference value. If the difference between the measured values in adjacent cycles is larger than the reference value, the measured values obtained in the new cycle are used as noise and the previous measured values are continuously adopted. At the same time, in the next cycle, the difference between the detected measurement value and the currently adopted measurement value is compared with a value that is twice the reference value, and if the detected measurement value is larger, the detected measurement value is compared in the next cycle. The difference between the measured value and the currently adopted measurement value is compared with a value multiplied by three times the reference value, and it is detected whether the noise is continuous in each period after the noise detection period, and the measurement detected in a certain period is performed. When the difference between the measured value and the currently adopted measured value becomes smaller than the product of the reference value and the total number of cycles from the noise detection cycle, the detected measured value is adopted and in subsequent cycles, adjacent cycles are used. A method for detecting a peak value of a measuring device output, which compares measured values and is characterized in that if the adopted measured value is larger than all measured values in a predetermined number of subsequent cycles, this measured value is taken as the peak value.