Noise removal filter and noise removal method

Abstract

Provided is a noise removal filter 1 for removing noise on an input signal, wherein: the noise removal filter 1 comprises a periodic noise removal filter 11 for removing periodic noise from the input signal, a high-frequency removal filter 12 for removing high-frequency noise from an output signal of the periodic noise removal 11 filter, and a stability assessment unit 13 for monitoring the stability of the output signal of the periodic noise removal filter 11; before the stability assessment unit 13 has determined that the output signal of the periodic noise removal filter 11 is stable, the output signal of the periodic noise removal filter 11 is outputted without being processed by the high-frequency removal filter 12; and after it is determined that the output signal of the periodic noise removal filter 11 is stable, the signal processed by the high-frequency removal filter 12 is outputted.

Description

Noise removal filter and noise removal method

[0001] The present invention relates to a noise removal filter and a noise removal method for removing noise contained in an input signal.

[0002] In various signal processing, a noise reduction filter is used to remove the influence of noise on a signal to be acquired. As an example of such a noise reduction filter, a technology relating to a digital filtering device and method for a signal from a linear solenoid is disclosed in Patent Document 1.

[0003] Japanese Patent Application Laid-Open No. 2006-295743

[0004] There are various types of noise removal filters depending on the type of noise to be removed. These filters include periodic noise removal filters for removing periodic noise and high-frequency removal filters for removing high-frequency noise. Periodic noise removal filters are used to remove noise of known frequencies, while high-frequency removal filters are used to remove high-frequency bands of randomly introduced noise. Conventionally, when using a periodic noise removal filter and a high-frequency removal filter, they are operated simultaneously on an input signal (i.e., an input signal that has already been passed through the periodic noise removal filter is passed through the high-frequency removal filter (or vice versa) from the start of processing), which results in a decrease in responsiveness due to these filter processes or a relatively long time required for the signal to stabilize.

[0005] In view of the above, an object of the present invention is to shorten the time until a signal becomes stable or to improve responsiveness in a noise removal filter or noise removal method that uses a periodic noise removal filter and a high frequency removal filter.

[0006] (Configuration 1) A noise removal filter for removing noise carried on an input signal, comprising: a periodic noise removal filter that removes periodic noise from the input signal; a high-frequency removal filter that removes high-frequency noise from an output signal of the periodic noise removal filter; and a stability determination unit that monitors the stability of the output signal of the periodic noise removal filter, wherein before the stability determination unit determines that the output signal of the periodic noise removal filter is stable, the output signal of the periodic noise removal filter is output without passing through the high-frequency removal filter, and after it is determined that the output signal of the periodic noise removal filter is stable, the noise removal filter outputs the signal that has passed through the high-frequency removal filter.

[0007] (Configuration 2) The noise removal filter according to Configuration 1, wherein, before it is determined that the output signal of the periodic noise removal filter is stable, the periodic noise removal filter is operated by sampling at a second cycle that is shorter than a sampling cycle used in a system that uses the value of the input signal.

[0008] (Configuration 3) The noise removal filter according to configuration 1 or 2, wherein the high frequency removal filter is reset at predetermined intervals, or the periodic noise removal filter and the high frequency removal filter are reset at predetermined intervals.

[0009] (Configuration 4) The noise removal filter according to configuration 3, wherein the predetermined period is the same as a sampling period used in a system that uses the value of the input signal.

[0010] (Configuration 5) The noise removal filter according to configuration 1 or 2, wherein when the stability determination unit determines that the output signal of the periodic noise removal filter has changed from a stable state to an unstable state, the high frequency removal filter is reset, or the periodic noise removal filter and the high frequency removal filter are reset.

[0011] (Configuration 6) A noise removal method for removing noise carried on an input signal, comprising: before it is determined that an output signal from a periodic noise removal filter that removes periodic noise from the input signal is stable, outputting an output signal from the periodic noise removal filter; and after it is determined that the output signal from the periodic noise removal filter is stable, outputting an output signal from a high frequency removal filter that removes high frequency noise from the output signal from the periodic noise removal filter.

[0012] (Configuration 7) The noise removal method according to Configuration 6, wherein, before it is determined that the output signal of the periodic noise removal filter is stable, the periodic noise removal filter is operated by sampling at a second cycle that is shorter than a sampling cycle used in a system that uses the value of the input signal.

[0013] (Configuration 8) The noise removal method according to configuration 6 or 7, wherein the high frequency removal filter is reset at predetermined intervals, or the periodic noise removal filter and the high frequency removal filter are reset at predetermined intervals.

[0014] (Configuration 9) The noise removal method according to configuration 8, wherein the predetermined period is the same as a sampling period used in a system that uses the value of the input signal.

[0015] (Configuration 10) The noise removal method according to Configuration 6 or 7, wherein, when it is determined that the output signal of the periodic noise removal filter has changed from a stable state to an unstable state, the high frequency removal filter is reset, or the periodic noise removal filter and the high frequency removal filter are reset.

[0016] According to the present invention, in a noise removal filter or noise removal method that uses a periodic noise removal filter and a high frequency removal filter, it is possible to shorten the time until a signal becomes stable or improve responsiveness.

[0017] FIG. 1 shows a conceptual configuration of a noise removal filter according to a first embodiment of the present invention. FIG. 2 shows a flowchart outlining the operation of the noise removal filter according to the first embodiment. FIG. 3 shows a graph explaining the function of the noise removal filter according to the first embodiment. FIG. 4 shows a conceptual configuration of a noise removal filter according to a second embodiment. FIG. 5 shows a flowchart outlining the operation of the noise removal filter according to the second embodiment. FIG. 6 shows a graph explaining the function of the noise removal filter according to the second embodiment. FIG. 7 shows a conceptual configuration of a noise removal filter according to a third embodiment.

[0018] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Note that the following embodiments are merely examples of how the present invention can be realized, and are not intended to limit the scope of the present invention.

[0019] <Embodiment 1> Fig. 1 is a block diagram showing the conceptual configuration of a noise removal filter according to embodiment 1 of the present invention. The noise removal filter 1 of this embodiment is a filter for removing the effects of noise carried on a sensor value (a digital signal input from an A / D conversion unit 2 that A / D converts the analog sensor value) from a temperature sensor when measuring the temperature of a controlled object in a temperature regulator. The noise removal filter 1 includes: a periodic noise removal filter 11 that removes periodic noise from an input signal from the A / D conversion unit 2; a high-frequency removal filter 12 that removes high-frequency noise from an output signal from the periodic noise removal filter 11; a stability determination unit 13 that monitors the stability of the output signal from the periodic noise removal filter 11; and a switch 14 that bypasses the high-frequency removal filter 12. For ease of illustration and understanding, Fig. 1 depicts each component of the noise removal filter 1 as if it were hardware. However, the noise removal filter 1 of this embodiment is a digital filter configured as software executed by a computing device such as a microcomputer (although, of course, each component or part of the component of the noise removal filter 1 may be configured as hardware (circuitry)).

[0020] The periodic noise removal filter 11 itself can be any noise filter capable of removing periodic noise, and the high frequency removal filter 12 itself can be any noise filter capable of removing high frequency noise.

[0021] The stability determination unit 13 determines whether the output signal of the periodic noise removal filter 11 is stable using a predetermined method. Before determining that the output signal of the periodic noise removal filter 11 is unstable, the stability determination unit 13 switches the switch 14 to bypass the high frequency removal filter 12, thereby outputting the output signal of the periodic noise removal filter without passing through the high frequency removal filter 12. On the other hand, after determining that the output signal of the periodic noise removal filter 11 is stable, the stability determination unit 13 switches the switch 14 to the high frequency removal filter 12 side, operates the high frequency removal filter 12, and outputs the signal that has passed through the high frequency removal filter 12. Note that the determination of signal stability is made based on sequentially sampled signal values, such as by determining whether a slope obtained from signal values ​​acquired within a predetermined time is within a predetermined range, or by determining whether a variation (value difference) in signal values ​​acquired within a predetermined time is within a predetermined range (this may be based on any method that can be used to determine signal stability, such as based on the characteristics of the target signal). As described above, in this embodiment, the noise removal filter 1 is configured as software, and therefore the description of "switching of the switch 14" is conceptual, and does not mean that the noise removal filter 1 has a hardware switch (however, as described above, the noise removal filter 1 may be configured as a circuit (hardware)).

[0022] Next, an outline of the operation of the noise removal filter (and an outline of the noise removal method) of this embodiment will be described with reference to FIG.

[0023] When a system (a temperature controller in this embodiment) using the noise removal filter issues an instruction to start operation of the noise removal filter, in step 201, the periodic noise removal filter 11 starts operating while sampling a signal at a second cycle. The "second cycle" is shorter than the "sampling cycle of a signal (a sensor value of a temperature sensor in this embodiment) used in the system (a temperature controller in this embodiment)." The "second cycle" is preferably set so that the periodic noise removal filter has a resolution that allows the periodic noise removal filter to remove periodic noise (to stabilize the output signal from the periodic noise removal filter) within the "system sampling cycle" (this is determined appropriately depending on the cycle of the target noise, the performance (specifications) of the periodic noise removal filter, etc.). Note that it is most preferable that the periodic noise removal filter be able to remove periodic noise (to stabilize the signal) within one "system sampling cycle." However, the periodic noise removal filter may be configured to remove noise within, for example, two "system sampling cycles" (this may be determined appropriately depending on the desired performance).

[0024] In step 202 following step 201, the stability determination unit 13 determines whether the output signal from the periodic noise removal filter is stable or not. If it is stable, the process proceeds to step 203; if it is not stable, the process proceeds to step 206.

[0025] In step 203, when it is determined that the signal is stable, the high frequency elimination filter 12 is operated to perform high frequency noise elimination processing on the output signal of the periodic noise elimination filter 11. In the following step 204, it is determined whether or not the first period has been reached, and if the first period has not been reached, the process returns to step 202 and the above processing is repeated until the first period is reached. On the other hand, if the first period is reached, the process proceeds to step 205 and the signal is output via the high frequency elimination filter 12 as described above. Note that the "first period" is the "sampling period of a signal (in this embodiment, the sensor value of the temperature sensor) used in the system (in this embodiment, the temperature controller)." In other words, the signal is output at the sampling period on the system side.

[0026] On the other hand, in step 206, when it is determined that the signal is not stable, it is determined whether or not the first period has been reached, and if the first period has not been reached, the process returns to step 202 and the above processing is repeated until the first period is reached, and if the first period is reached, the process proceeds to step 207, where the high frequency elimination filter 12 is bypassed as described above and the signal is output from the periodic noise elimination filter 11. Note that if the system is configured so that "the periodic noise elimination filter can remove periodic noise within one period of the 'system sampling period' (the signal is stable)," as described above, the processing of steps 206-207 is not particularly necessary (in embodiments 2 and 3 described later, the processing of steps 206-207 is not required).

[0027] In step 208, it is determined whether the processing has ended (whether or not there has been an end instruction from the system, etc.), and if it has not ended, the processing returns to step 201 and the above processing is repeated (which causes a signal to be repeatedly output in the first cycle), and if there has been an end instruction, etc., the processing ends.

[0028] FIG. 9 shows graphs illustrating an example of the operation of a conventional noise reduction filter (not including the configuration of the present invention) equipped with a periodic noise reduction filter and a high-frequency reduction filter. In FIG. 9( a), signal N represents noise, TS represents the desired signal value (assumed to be zero output for ease of explanation), and SS represents the signal value acquired in the first period (the system sampling period: 30 ms in this example). FIG. 9( b) shows the signal obtained by filtering signal SS (a sensor value containing noise) through the periodic noise reduction filter and the high-frequency reduction filter, F1S represents the output obtained by filtering signal SS through the periodic noise reduction filter, and F2S represents the output obtained by filtering the output of the periodic noise reduction filter (F1S) through the high-frequency reduction filter. As shown in FIG. 9( b), in the conventional method in which the periodic noise reduction filter and the high-frequency reduction filter are operated simultaneously from the beginning, it takes 14 periods (420 ms) before the final output, F2S, can be considered stable. This is because it takes multiple samples for the output of the periodic noise removal filter to stabilize, and because operating the high frequency removal filter on the output of the unstable periodic noise removal filter takes longer for the output of the high frequency removal filter to stabilize.

[0029] In contrast, according to the noise removal filter of this embodiment, by sampling at the second period and operating the periodic noise removal filter, the operation of the periodic noise removal filter can be stabilized early, and by operating the high-frequency removal filter after the output of the periodic noise removal filter has stabilized, it is possible to prevent the output of the high-frequency removal filter from taking a long time to stabilize due to the influence of the unstable output of the periodic noise removal filter. Figure 3 is a graph illustrating an example of the operation of the noise removal filter of this embodiment (the periodic noise removal filter itself and the high-frequency removal filter themselves are the same as those of the conventional example in Figure 9, but the method of this embodiment is applied). As in Figure 7, N represents noise, F1S represents the output obtained by applying the periodic noise removal filter to the input signal (sampled at the second period), and F2S represents the output obtained by applying the high-frequency removal filter to the output of the periodic noise removal filter (F1S). 3, sampling at the second period (0.6 ms in this example) allows the operation of the periodic noise removal filter to be stabilized early, and the output of the high frequency removal filter is also stabilized early by operating the high frequency removal filter with a delay (after the signal has stabilized). As a result, a stable output is obtained before the arrival of the first period (30 ms).

[0030] As described above, according to this embodiment, in a noise removal filter or noise removal method using a periodic noise removal filter and a high frequency removal filter, it is possible to improve the reduction in the time until a signal becomes stable.

[0031] While the present embodiment exemplifies continuous sampling at the "second period," the present invention is not limited to this. It is sufficient to perform sampling at the "second period" at least "before it is determined that the output signal of the periodic noise removal filter is stable." After it is determined that the output signal of the periodic noise removal filter is stable, sampling may be performed, for example, at the "sampling period used in the system." While sampling at the "second period" is preferable, sampling at the "second period" is not essential to the present invention. Sampling at the "sampling period used in the system" may be performed as in the past. In this case, the effect of "the signal is stable at the time of the first sampling (first period)" as in the present embodiment cannot be obtained, but the effect of "reducing the time it takes for the output of the high frequency removal filter to stabilize due to the influence of the output of the unstable periodic noise removal filter" can be obtained.

[0032] <Embodiment 2> Fig. 4 is a block diagram showing the conceptual configuration of a noise removal filter according to embodiment 2 of the present invention. Similar to noise removal filter 1 of embodiment 1, noise removal filter 1-2 of this embodiment is a filter for removing noise present in the sensor value from the temperature sensor when sampling the temperature of the controlled object in a temperature regulator. The same components as those in embodiment 1 (Fig. 1) are designated by the same reference numerals as those in embodiment 1, and their description here will be simplified or omitted.

[0033] The noise removal filter 1-2 of this embodiment has a reset unit 15 that resets the periodic noise removal filter 11 and the high frequency removal filter 12 (restarts the filtering process) at predetermined intervals. In this embodiment, the "predetermined interval" is the same as the first interval (the "signal sampling interval used in the system") described in the first embodiment. Note that the noise removal filter 1-2 of this embodiment is configured as software, as in the first embodiment, and the reset unit 15 is configured as a part of it (as software); however, the reset unit may also be configured as hardware.

[0034] Next, an outline of the operation of the noise removal filter (and an outline of the noise removal method) of this embodiment will be described with reference to Fig. 5. Note that the same processing concepts as those in the first embodiment will be designated by the same reference numerals as those in the first embodiment (Fig. 2), and the description here will be simplified or omitted.

[0035] Steps 201 to 208 are the same processes as those described in embodiment 1. In step 204, it is determined that the first period has arrived, and in step 205, the filtered signal is output in the first period. In step 501, the periodic noise removal filter 11 and the high frequency removal filter 12 are reset (that is, in the filtering process, an operation is performed such that the filter is not affected by the previous sampled value for each first period).

[0036] 9(c) is a graph illustrating an example of the behavior of a conventional noise removal filter (not having the configuration of the present invention) equipped with a periodic noise removal filter and a high frequency removal filter when TS (the signal value that is originally desired to be obtained) changes stepwise. As in the description of embodiment 1, N represents noise, and F2S represents the output after passing through the periodic noise removal filter and the high frequency removal filter. As shown in FIG. 9(c), the response to the change in TS is delayed due to the effects of filtering.

[0037] In contrast, with the noise removal filter of this embodiment, the periodic noise removal filter 11 and the high-frequency removal filter 12 are reset at the first cycle, thereby reducing the effect of the signal value before the change, which can lead to a time lag before the signal value changes (delay in the response of the filtered output value). Figure 6 is a graph illustrating an example of the operation of the noise removal filter of this embodiment (the periodic noise removal filter itself and the high-frequency removal filter themselves are the same as those of the conventional example shown in Figure 9 , but with the method of this embodiment applied). N represents noise, F1S represents the output obtained by filtering the input signal (sampled at the second cycle) with the periodic noise removal filter, and F2S represents the output obtained by filtering the output (F1S) of the periodic noise removal filter. In this embodiment, the processing described in the first embodiment is repeated every first cycle (30 ms, as in the first embodiment), resulting in processing that is not affected by the sampled value in the previous cycle. As a result, as can be seen from Figure 6 , high responsiveness to changes in TS is achieved.

[0038] As described above, according to this embodiment, in a noise removal filter or noise removal method that uses a periodic noise removal filter and a high frequency removal filter, it is possible to improve the shortening of the time until a signal becomes stable, as in the first embodiment, and also to improve responsiveness.

[0039] While the present embodiment exemplifies a case in which the periodic noise elimination filter and the high-frequency elimination filter are reset at a first cycle (the "sampling cycle of the signal used in the system"), the present invention is not limited to this example and may be reset at any cycle. However, resetting at a cycle shorter than the first cycle is not very meaningful, and resetting at a cycle longer than the response delay time shown in FIG. 9C (which varies depending on the type of target signal, the specifications of the filter used, etc.) is also not very meaningful. Therefore, the reset cycle should be equal to or longer than the first cycle and shorter than the response delay time. Furthermore, while the present embodiment exemplifies a case in which both the periodic noise elimination filter and the high-frequency elimination filter are reset, the present invention is not limited to this example and may be reset either the periodic noise elimination filter or the high-frequency elimination filter (if only one of the filters is reset, it is more effective to reset the high-frequency elimination filter, which takes longer to process).

[0040] <Embodiment 3> Fig. 7 is a block diagram showing the conceptual configuration of a noise removal filter according to embodiment 3 of the present invention. Similar to noise removal filter 1 of embodiment 1, noise removal filter 1-3 of this embodiment is a filter for removing noise present in the sensor value from the temperature sensor when sampling the temperature of the controlled object in a temperature regulator. The same components as those in embodiment 1 (Fig. 1) are designated by the same reference numerals as those in embodiment 1, and their description here will be simplified or omitted.

[0041] The noise removal filter 1-3 of this embodiment includes a stability determination and reset unit 16 that has the reset function described in embodiment 2 in the stability determination unit described in embodiment 1. The stability determination and reset unit 16 resets the periodic noise removal filter 11 and the high frequency removal filter 12 when it is determined that the output signal of the periodic noise removal filter 11 has changed from a stable state to an unstable state. Note that the noise removal filter 1-3 of this embodiment is configured as software as in embodiment 1, and the stability determination and reset unit 16 is configured as a part thereof (as software), but it may also be configured as hardware.

[0042] Next, an outline of the operation of the noise removal filter (and an outline of the noise removal method) of this embodiment will be described with reference to Fig. 8. Note that the same processing concepts as those in the first embodiment will be designated by the same reference numerals as those in the first embodiment (Fig. 2), and the description here will be simplified or omitted.

[0043] Steps 201 to 203 are the same as those described in the first embodiment. If it is determined in step 202 that the output signal of the periodic noise removal filter 11 is stable and then it is determined that the output signal of the periodic noise removal filter 11 has changed from a stable state to an unstable state, the periodic noise removal filter 11 and the high frequency removal filter 12 are reset, and the process is restarted from the beginning (step 801: Yes → step 802 → step 201). On the other hand, if the output signal of the periodic noise removal filter 11 remains stable, the process from step 204 onwards (the same process as that described in the first embodiment) continues. Note that the determination of stability or instability may use the same threshold value as in step 202 (threshold value related to the "slope" or "difference" described in the first embodiment), or may use a threshold value different from that in step 202 (for example, a threshold value that determines "unstable" when there is a larger fluctuation).

[0044] When the signal value that is originally desired to be obtained (in this embodiment, the original sensor value of the temperature sensor) changes, the output signal of the periodic noise removal filter also fluctuates accordingly. Based on this fluctuation, an "unstable state" is determined, and the periodic noise removal filter 11 and the high frequency removal filter 12 are reset (i.e., after the signal value changes, the operation is such that the signal value is no longer affected by the value before the change). This reduces the time it takes to reach the changed signal value (delayed response) due to the influence of the signal value before the change.

[0045] As described above, according to this embodiment, in a noise removal filter or noise removal method that uses a periodic noise removal filter and a high frequency removal filter, it is possible to improve the shortening of the time until a signal stabilizes, as in the first embodiment, and also to improve responsiveness, as in the second embodiment.

[0046] In this embodiment, an example is given in which both the periodic noise removal filter and the high frequency removal filter are reset, but the present invention is not limited to this, and it is also possible to reset either the periodic noise removal filter or the high frequency removal filter (if only one of the filters is to be reset, it is more effective to reset the high frequency removal filter, which takes longer to process).

[0047] In each embodiment, the present invention has been described as a noise removal filter (or a noise removal method) for removing noise from a sensor value (input signal) from a temperature sensor in a temperature controller, but the present invention is not limited to this and can be used as a noise removal filter (or a noise removal method) for any signal.

[0048] 1. Noise removal filter 11. Periodic noise removal filter 12. High frequency removal filter 13. Stability determination section

Claims

1. A noise removal filter for removing noise carried on an input signal, comprising: a periodic noise removal filter for removing periodic noise from the input signal; a high-frequency removal filter for removing high-frequency noise from the output signal of the periodic noise removal filter; and a stability determination unit for monitoring the stability of the output signal of the periodic noise removal filter. Before it is determined by the stability determination unit that the output signal of the periodic noise removal filter is stable, the output signal of the periodic noise removal filter is output without passing through the high-frequency removal filter, and after it is determined that the output signal of the periodic noise removal filter is stable, a signal passing through the high-frequency removal filter is output.

2. The noise removal filter according to claim 1, wherein the periodic noise removal filter is operated by sampling at a second period that is shorter than the sampling period used in a system that uses the value of the input signal before it is determined that the output signal of the periodic noise removal filter is stable.

3. The noise removal filter according to claim 1 or 2, wherein the high-frequency removal filter is reset at a predetermined period, or the periodic noise removal filter and the high-frequency removal filter are reset at a predetermined period.

4. The noise removal filter according to claim 3, wherein the predetermined period is the same as the sampling period used in a system that uses the value of the input signal.

5. The noise removal filter according to claim 2, wherein when it is determined by the stability determination unit that the output signal of the periodic noise removal filter has changed from a stable state to an unstable state, the high-frequency removal filter is reset or the periodic noise removal filter and the high-frequency removal filter are reset.

6. A noise removal method for removing noise carried on an input signal, wherein before it is determined that the output signal from a periodic noise removal filter for removing periodic noise from the input signal is stable, the output signal of the periodic noise removal filter is output, and after it is determined that the output signal of the periodic noise removal filter is stable, an output signal from a high-frequency removal filter for removing high-frequency noise from the output signal of the periodic noise removal filter is output.

7. The noise removal method according to claim 6, wherein the periodic noise removal filter is operated by sampling at a second period that is shorter than the sampling period used in the system that uses the value of the input signal before it is determined that the output signal of the periodic noise removal filter is stable.

8. The noise removal method according to claim 6 or 7, wherein the high-pass filter is reset at a predetermined period, or the periodic noise removal filter and the high-pass filter are reset at a predetermined period.

9. The noise removal method according to claim 8, wherein the predetermined period is the same as the sampling period used in the system that uses the value of the input signal.

10. The noise removal method according to claim 7, wherein when it is determined that the output signal of the periodic noise removal filter has changed from a stable state to an unstable state, the high-pass filter is reset or the periodic noise removal filter and the high-pass filter are reset.

Images