JP5485823B2 - Inverter control circuit - Google Patents

Description

  Embodiments described herein relate generally to a control circuit for an inverter device that is mounted on an inverter device that performs variable speed control of an electric motor and that reads and controls a signal applied to an input terminal.

  The inverter device controls the voltage and frequency output to the motor in order to perform variable speed control of the motor. Such an inverter device is often installed in an environment exposed to noise generated by rotating an electric motor, noise generated by operating other devices, and the like. Then, noise is applied to the input terminal of the inverter device, and a change in the input state may be erroneously detected.

  In general, in order to avoid the influence of noise as described above, a noise filter is arranged at an input terminal, or a filter time for filtering an input signal by software is set long. For example, when the level of the input terminal is read out three times every 1 μs and the majority result is regarded as a valid data value, this processing is further performed a predetermined number of times every 1 ms, and all data values are the same level. In addition, a process of determining and updating the data value given to the input terminal is performed.

JP 7-212209 A

  However, in the noise filter, the setting of the filter constant (cutoff frequency) does not always match the characteristics of the noise that is actually generated. In addition, if the filter time is set longer with a margin, there is a problem that the input response is delayed. In the case of the filtering process by software as described above, it is effective for randomly generated noise.

However, for example, when a relay or contactor used in a system used by a user in the field is controlled to be turned ON / OFF, periodic noise may occur. For example, if noise with a pulse width of 3 μsec or more occurs in a cycle of 1 msec, the above filtering process cannot eliminate the noise, and the sampled noise level may be accepted as data, resulting in malfunction. . Such noise cannot be removed even if the number of processes performed every 1 ms is increased.
Therefore, a control circuit for an inverter device that can appropriately avoid the influence of noise according to the characteristics of noise that is actually generated is provided.

  According to the embodiment, in the control circuit that is mounted on the inverter device that performs variable speed control of the electric motor and performs the control by reading the signal given to the input terminal, the sampling means keeps the binary level signal given to the input terminal constant. Sampling is performed for each cycle, and the sampled data value is stored in the storage means. The noise information detection means, when the operation by the sampling means elapses for a predetermined time, the data value “0” for the data value stored in the storage means. ”And the length of the period in which the data value“ 1 ”is continuous, the noise generation cycle Tw estimated to be a change in the data value due to the occurrence of noise and the state in which the data value has changed are continuous. The noise duration Th to be detected is detected. Then, the sampling setting means, based on the noise generation period Tw and the noise duration time Th, the period Tin and / or the sampling that the sampling means samples data during normal processing so as to avoid the influence of noise. The number of times Rn for sampling data continuously for each period is set.

The flowchart which is 1st Embodiment and shows a noise detection process Noise detection processing timing chart Flowchart showing processing for determining the reading cycle Tin and the number of continuous readings Rn Flow chart showing input signal processing The figure which shows the processing image of FIG. Flow chart showing processing for changing the cycle Tin and the number of times Rn Functional block diagram schematically showing the configuration of the inverter device The figure which shows an example of the data which performed frequency distribution processing Diagram showing noise data example A The figure which shows the analysis result of noise data example A Diagram showing noise data example B The figure which shows the analysis result of noise data example B Figure showing a histogram Diagram showing noise data example C The figure which shows the analysis result of noise data example C Flow chart showing noise detection request activation processing The figure which is 2nd Embodiment and shows the example of an output of alarm data FIG. 7 equivalent view showing the third embodiment Diagram showing an example of noise data used for verification

(First embodiment)
Hereinafter, a first embodiment will be described with reference to FIGS. FIG. 7 is a functional block diagram schematically showing the configuration of the inverter device. The inverter device 1 includes a control circuit 2 and an inverter main circuit 3, and a signal supplied to the input terminal 4 is supplied to the control circuit 2 via the input circuit 5. The input circuit 5 includes, for example, a low-pass filter and a protection circuit. The control circuit 2 is constituted by a CPU or a microcomputer, and gives a control signal to the inverter main circuit 3 in accordance with a signal given to the input terminal 4 to control driving of the electric motor 6 composed of, for example, a three-phase induction motor.

  The control circuit 2 includes a calculation unit 7, an input data storage / save unit 8, an input data analysis result storage unit 9, a continuous data acquisition interval storage unit 10, a sampling cycle storage unit 11, an input data ON determination time storage unit 12, An input data OFF determination time storage unit 13, an input terminal data collection time storage unit 14, a deviation value determination unit 15, an alarm output unit 16, and the like are provided. Each of the storage units 8 to 14 is a memory, and the deviation value determination unit 15 is externally shown as a software functional part executed by the calculation unit 7.

  Next, the operation of this embodiment will be described. FIG. 4 is a flowchart showing input signal processing generally performed by the control circuit 2. This process is repeatedly executed by a timer interrupt or the like that occurs every cycle Tin (for example, Tin = 500 μsec) for sampling the data of the input terminal 4. Basically, the data of the input terminal 4 is read every 1 microsecond, and it is determined that the data when it is continuously at the same level n (≧ 3) times is valid (majority processing). The 1 μs is stored in the continuous data acquisition interval storage unit 10 in advance and the sampling period Tin is stored in the sampling cycle storage unit 11 in advance. For “n” in the majority process, the data value is “1”. The input data ON determination time storage unit 12 stores the data value “0” in the input data OFF determination time storage unit 13.

First, when the input counter is cleared to zero and initialized by substituting "input terminal valid data" for "previous input terminal valid data" (S300), the data value sampled from the current input terminal 4 for "previous data" Is substituted (S301). Then, dummy delay processing is performed so that the period when data is continuously read is 1 μsec (S302).
Next, when the data value sampled from the input terminal 4 is substituted for “new data” (S310), it is determined whether or not “previous data” and “new data” match (S311). (YES) If S3111 does not match, the process proceeds to (NO) step S3115. In step S3111, the “input counter” is incremented, and subsequently, “new data” is substituted into “previous data” for updating (S3112).

  Next, it is determined whether or not the count value of the “input counter” is “n” (S3113). If it is “n” (YES), “input terminal valid data” is updated with “new data” (S3114). If not "n" (NO), the process returns to step S310. Here, the time required for the processing of steps S311 to S3113 is 1 microsecond. In step S 3115 (A), if “input terminal valid data” and “previous input terminal valid data” are the same (YES), “continuous valid count” is incremented (B), and if they are different (NO) “ “Continuous valid count” is reset (C).

  “Continuous valid count” counts the number of times the same value continues as “input terminal valid data”. In the input terminal processing of the inverter device 1, ON / OFF (1/0) data of the input terminal 4 Processing is performed using both of the measurement time (for example, an operation command is validated when ON continues for 5 ms). After executing step S3115, the process ends. If “input terminal valid data” is not updated in step S3114, the previous “input terminal valid data” remains unchanged. FIG. 5 conceptually illustrates the above processing.

In performing the input signal processing as described above, when periodic noise is generated and superimposed, the processing for detecting the noise generation cycle and the duration in which the noise data is continuously generated, This will be described below.
FIG. 1 is a flowchart showing noise detection processing. This process is started when a noise detection process is requested (see FIG. 16 to be described later), and once called, the process is repeated every 1 μs, and ends when the condition is satisfied. First, “0” is assigned to “previous data”, “time”, “A time”, and “B time”, respectively, and a pointer of a table storing each data is set to an initial value (S1). Then, when “time” is incremented (corresponding to 1 μsec) (S2), the data value sampled from the input terminal 4 is substituted for “new data” (S3). Then, it is determined whether or not “previous data” and “new data” match (S4). If both match and there is no change (YES), for example, a NOP command is used to set the processing time to 1 μsec. After performing a dummy process (S5) such as executing, the process proceeds to step S7. In this case, only the time is updated.

  On the other hand, if “previous data” and “new data” have changed without matching in step S4 (NO), it is determined in subsequent step S6 whether the value of “new data” is ON, that is, “1”. To do. If the value of “new data” is ON (YES), the data value has changed from “0” to “1”. In this case, “time” is substituted into “A time (OFF time)”. Store (S610), and then perform a dummy process similar to step S5 (S611).

  In step S6, if the value of “new data” is not ON (NO), the data value has changed from “1” to “0”. In this case, “time A” is subtracted from “time”. Is stored in “B time (ON time)” (S620). Further, “time” is substituted for “noise interval” and stored (S621). When ON time and noise interval data are stored in the storage table (S622), the storage table pointer is updated (increment; S623). “Time” is cleared to zero (S624). The relationship between “A time”, “B time”, “noise interval”, etc. is shown in FIG. Thereafter, dummy processing similar to steps S5 and S611 is performed (S625).

  When “new data” is set in “previous data” (S7), the completion of the data input process is confirmed in the subsequent step S8. Here, it is determined whether or not the pointer has reached “500”, or whether or not a predetermined time has elapsed. If the pointer has not been completed, the process returns to step S2. With this process, it is possible to collect 500 times of noise period and ON time (noise duration Th) data. The sampled data is stored in the input data storage / save unit 8. Here, how much data is sampled (collected) and the predetermined time are stored in the input terminal data collection period storage unit 14 in advance.

  Note that when the predetermined time has elapsed and the number of samples is less than “500”, there is no problem because the number of noise occurrences is small. As the predetermined time, for example, if the processing cycle of a normal input terminal is about 500 μs, about 500 ms is sufficient if it is twice that (500 times). When the number is less than 500, only noise width countermeasures after S412 shown in FIG.

  Next, processing (noise analysis processing) that analyzes the data collected by the above processing and reflects the signal processing at the input terminal 4 in the sampling period Tin and the continuous sampling count Rn is shown in the flowchart of FIG. This process is executed only when it is determined that noise is occurring (S400: NO). That is, as a result of analyzing the collected data, if no data change occurs, it is determined that no noise is generated (YES).

In the following, assuming that the ON time is Ton, first, an average value Tonave and standard deviation Tonσ are obtained for the ON time Ton, and a frequency distribution table is created (S401). When the standard deviation Tonσ is small and has a certain width, there is a high possibility of erroneous recognition due to the influence of noise. Further, when obtaining the frequency distribution, it is considered that there is a corresponding number of data at the center value of the data range, and the average value Tonave and the standard deviation Tonσ are obtained. These analysis results are stored in the input data analysis result storage unit 9.
Next, the average value Twave and the standard deviation Twσ are similarly obtained for the noise period Tw, and a frequency distribution table is created (S402). When the standard deviation Twσ is small and there is a relationship that is an integral multiple of the sampling period Tin of the input data, there is a high possibility that erroneous recognition due to the influence of noise will occur. When obtaining the frequency distribution, it is considered that there is a corresponding number of data at the center value of the data range, and the average value Twave and the standard deviation Twσ are obtained.

In a succeeding step S460, it is determined whether or not a value obtained by adding five times the standard deviation Twσ to the average value Twave is within a range exceeding 400 μsec and less than 1 msec. That is, it is determined whether or not the noise generation period Tw is in the vicinity of the sampling period Tin (within a predetermined range including the period Tin). If “YES” here, the process proceeds to a step S461, and if “NO”, the process proceeds to the step S410. When the process proceeds to step S461, it indicates that the intervals of the noise period Tw are within a certain range, so the sampling period Tin is set by the equation (1). The right side is equal to the value used in step S460.
Tin = Twave + Twσ × 5 (1)

  In step S410, it is determined whether or not the standard deviation Tonσ is less than 1 μsec. When Tonσ <1 μs (YES), it indicates that noise is concentrated on the average value Tonave ± 3 μm. In this case, the process proceeds to step S411. If Tonσ ≧ 1 μs (NO), the process proceeds to step S420. In step S411, it is determined whether or not the average value Tonave is less than 3 μs. In the case of Tonave <3 μs (YES), it is possible to eliminate the influence of noise by setting the number Rn of continuous reading of data at the input terminal as in step S412 and subsequent steps. On the other hand, if Tonave ≧ 3 μsec, the process proceeds to step S420.

In step S412, the number of continuous samplings Rn is set by equation (2).
Rn = Tonave + Tonσ × 5 (2)
In this case, only the number of times Rn may be changed. For example, when Tonave = 3 μ and Tonσ = 1 μsec, Rn = 8. When the noise alarm is set (S413), in the subsequent step S414, the continuous sampling count Rn is set to “3” or more (Rn ≧ n) as the minimum number of majority in the input signal processing. Therefore, if Rn <3 (A: YES), Rn = 3 is set (B) and the process is terminated.

  Next, the processing after Step S420 (when Tonσ ≧ 1 μs and Tonave ≧ 3 μs) will be described. First, it is determined whether or not the standard deviation Twσ is less than 50 μs (S420). If Twσ ≧ 50 μsec (NO), it is determined that the variance of the noise is large, and processing after step S441, which is analysis processing of the frequency distribution result, is executed. On the other hand, if Twσ <50 μsec (YES), the magnitude relationship between the current sampling period Tin and the average value Twave is determined (step S421). If Tin> Twave (YES), the sampling cycle Tin is changed (S422).

Hereinafter, the process performed in step S422 will be described. When the sampling period Tin matches an integer multiple of the noise period Tw, there is a high possibility of being affected by noise when sampling the data at the input terminal 4. Therefore, when the first decimal place of Tin / Twave is any one of “8, 9, 0, 1, 2”, it is determined that the relationship between the two is close to an integral multiple. That is, the following modulo operation is performed,
AA = MOD (Tin * 10 / Twave, 10) (3)
If the value of AA is any one of “8, 9, 0, 1, 2”, the sampling period Tin is changed. For example, in order to change the remainder to be “5”, the following calculation is performed.
BB = {INT (Tin * 10 / Twave) +5} * Twave / 10 (4)

Some specific numerical examples are given below.
<Twave = 260, Tin = 500>
AA = MOD (Tin * 10 / Twave, 10)
= MOD (5000 / 260,10) = 9
Therefore, the sampling period Tin is changed.
BB = {INT (Tin * 10 / Twave) +5) * Twave / 10
= {INT (5000/260) +5) * 260/10 = 624
In this case, the sampling period Tin is changed to 624 μsec.

<Twave = 240, Tin = 500>
AA = MOD (Tin * 10 / Twave, 10)
= MOD (5000 / 240,10) = 0
Also in this case, the sampling period Tin is changed.
BB = (INT (Tin * 10 / Twave) +5) * Twave / 10
= (INT (5000/240) +5) * 240/10 = 600
In this case, the sampling period Tin is changed to 600 μsec.

<Twave = 300, Tin = 500>
AA = MOD (Tin * 10 / Twave, 10)
= MOD (5000 / 300,10) = 6
In this case, it is determined that no change is necessary.

<Twave = 26, Tin = 500>
AA = MOD (Tin * 10 / Twave, 10)
= MOD (5000 / 26,10) = 2
Change the sampling period Tin.
BB = (INT (Tin * 10 / Twave) +5) * Twave / 10
= (INT (5000/26) +5) * 26/10 = 512
In this case, the sampling period Tin is changed to 512 μs.

<Twave = 30, Tin = 500>
AA = MOD (Tin * 10 / Twave, 10)
= MOD (5000 / 30,10) = 6
It is determined that no change is necessary. After execution of step S422, a noise alarm is set (S423).

  If Tin ≦ Twave in step S421 (NO), the processing in steps S430 to S433 is performed. The processes in steps S430 and S431 are the same as those in steps S412 and S414. In a succeeding step S432, it is determined whether or not Rn> 20. If Rn> 20, the influence of noise cannot be ignored, a noise alarm is set (S433), Tin and Rn are not changed, and the processing returns as it is (that is, the processing in FIG. 3 is terminated).

  Next, frequency distribution data processing after step S441 will be described. The frequency distribution is obtained by dividing the noise period data by a window (class) having a width of 50 μsec. FIG. 8 shows an example of data subjected to frequency distribution processing. If there is a frequency of 100 degrees or more (determined that a frequency that is twice or more in 10 windows is significant for 500 samplings) (S441: YES), the process proceeds to the next step S450. . If there is no frequency of 100 degrees or more, (NO) it is determined that there is no influence of noise, and the number of continuous samplings Rn is set after step S430, but the noise data does not show a significant feature. Since this is not necessarily correct, a noise alarm is set in step S451 before that.

  In step S450, if there is one noise cycle with a frequency of 100 degrees or more (YES), the center value of the window range is set to Twave, and the process of step S422 is executed. On the other hand, when there are a plurality of noise periods having a frequency of 100 degrees or more (NO), it is not easy to avoid the influence of noise by changing the sampling period Tin, and therefore the number of consecutive samplings Rn is set after step S430. In this case as well, since there are two or more noise periods of 100 degrees or more and the setting of Rn is not always correct, a noise alarm is set in step S451.

  In the above processing, the setting of the sampling cycle Tin and the number of continuous samplings Rn is limited to simple calculation. However, if advanced calculation is possible, it is based on advanced analysis such as data spectrum analysis. Based on the analysis results from these measured data, it is possible to set Tin and Rn that are not affected by noise.

  As a result of the above processing, processing for changing the sampling period Tin and the continuous sampling count Rn will be described with reference to FIG. The result of the noise diagnosis is that the data value of the input terminal 4 is determined not to change when the data value “0; OFF” or “1; ON” of the input terminal 4 continues for a specific time (for example, 1 second). The sampling period Tin and the number of readings Rn are changed. FIG. 6 shows a flowchart of the change process.

  First, it is determined whether there is a change request due to the influence of noise (step S11). The change request is reset by the initial setting process when the power is turned on, and is set when the processes of steps S412, S422, S430, S433, and S451 are executed. When there is a change request (YES), it is determined whether or not the state in which the data value of the input terminal 4 does not change has exceeded a certain time (step S12). The object to be changed based on, that is, the sampling cycle Tin or the continuous sampling count Rn is changed (step S13). Then, the change request is reset (step S14).

Hereinafter, the result of applying the above processing will be described for a specific example of noise data.
<Noise data example A>
In FIG. 9, “15” of No. 1 indicates that the state of the data value “1” (input terminal ON) continues 15 times every 1 μsec, and “998” of No. 2 indicates the data value. This indicates that the state of “0” (input terminal OFF) continues for 998 times.
When the data reaches a predetermined number (for example, 1000), the data is analyzed. No. 1 and No. 2, no. 3 and no. 4, that is, an interval in which the data value changes from “0” to “1” by pairing the odd number and the even number, and a width in which the state of the data value “1” continues in the odd number data. obtain. For example, for the data as shown in FIG. 8, it is assumed that the results of obtaining the average of the noise generation period Tw and the on-time Ton and the standard deviation Twσ are as shown in FIG.

Thus, since the average value Twave of the noise generation period Tw is 999 μs and the standard deviation Twσ is 28.8 μs, (Twave + Twσ × 5) is calculated to be 1143 μs, and since Tin> Twave, in step S421. It judges "YES" and performs Step S422.
AA = MOD (10000 / 999,10) = 0
BB = {INT (Tin * 10 / Twave) +5) * Twave / 10
= {INT (10000/999) +5) * 999/10 = 1498
In this case, the sampling period Tin is changed to 1498 μsec.

<Noise data example B>
FIG. 11 shows a different data example B. Similarly, when the analysis is performed as shown in FIG. 12, the standard deviation becomes relatively large as 245.4 μsec with respect to the average value 754 μsec of the noise generation period Tw. This is the case. In this case, when the frequency distribution of these data is obtained, it can be seen that noise is generated in two periods of 500 μsec and 1000 μsec, as shown in FIG.
In this case, when (Twave + Twσ × 5) is obtained, it is 1981 μs, “NO” is determined in the step S420, and the frequency distribution processing from the step S440 onward is performed. As shown in FIG. 13, since there are two times when the frequency is “100” or more, “NO” is determined in step S450, and the continuous sampling count Rn is set in step S430.

<Noise data example C>
14 and 15 are diagrams corresponding to FIGS. 10 and 11 of the noise data example C. FIG. In this case, when (Twave + Twσ × 5) is obtained, it is 1003.5 μsec, and Tonσ = 0.8 μsec <1 μsec and Tonave = 1.7 μsec <3 μsec, so the process of step S412 is executed.
Tonave + Tonσ × 5 = 1.7 + 0.8 × 5 = 5.7
→ n = 6
It becomes. Therefore, the number of continuous samplings Rn = 6 is set with the sampling period Tin unchanged.

Here, in the processing of FIG. 1, if data is accumulated for 500 ms every 1 μs in order to monitor noise at the input terminal 4, the number of necessary data bits is 500 ms / 1 μs = 500,000. When this is converted into the capacity of a memory for storing data in units of 1 byte, 500000/8 = 62.5 (k bytes) is required. Further, since it can be said that the noise analysis frequency of 500 degrees is sufficiently reliable, the required memory capacity is 500 × 2 × 2 = 2000, that is, 2 kbytes is sufficient.

  The reason why the noise period exceeds 2 bytes is 65.536 mS in the unit of 1 μsec, which is sufficient. Further, since the frequency can be made constant as 500 degrees regardless of the noise period, there is an advantage that the analysis is not affected by the noise period. The above noise data examples A to C can also be analyzed with data of 500 degrees. Further, as shown in FIG. 8, if the method stores the frequency for each width of 50 μs, the memory capacity can be extremely reduced to 10 × 2 × 1 = 20 (bytes).

  Further, when the inverter device 1 is actually used, a command is not frequently input via the input terminal 4. Therefore, the above noise diagnosis can be performed during the normal operation of the inverter device 1 when the processing capacity of the CPU used in the control circuit 2 is high. However, if the processing capacity of the CPU is not so high, or if it is necessary to prioritize processing of other functions, this noise diagnosis is performed after the power is turned on to output an alarm, sampling cycle Tin of the input terminal, and so on. The number of samplings Rn is set.

  The noise detection process (noise data sampling) shown in FIG. 1 is performed, for example, in a state where the inverter device 1 is not operating. Here, the “state where the inverter is not operating” means immediately after the inverter device 1 is turned on or immediately after being reset, before starting the operation of the inverter device 1 or after completing the operation stop. Thereby, the noise detection process can be performed without reducing the operation efficiency and the processing performance of the inverter device 1. Further, the noise detection process may be periodically performed regardless of the operation state of the inverter device 1.

  FIG. 16 is a flowchart for starting the noise detection process as described above. In step S21, it is determined whether or not there is a noise detection request that is set immediately after turning on the power of the inverter device 1, immediately after resetting, before the start of operation, after the completion of the operation stop, or periodically set. (YES) The noise detection process shown in FIG. 1 is executed (step S22). Then, the noise detection request is set to “none” (step S23).

  As described above, according to the present embodiment, in the control circuit 2 that is mounted on the inverter device 1 that performs variable speed control of the electric motor 6 and performs control by reading a signal given to the input terminal 4, The binary level signal given to 4 is sampled at regular intervals, and the sampled data value is stored in the input data storage / save unit 8. When the sampling operation has passed for a predetermined time, the data value stored in the input data storage / save unit 8 is determined by the noise from the length of the period in which the data value “0” continues and the length of the period in which the data value “1” continues. A noise generation period Tw estimated to be a change in the data value due to the occurrence of the noise and a noise duration Ton in which the state in which the data value has changed (state from 0 to 1) continues are detected.

  Then, in order to avoid the influence of noise based on the noise generation period Tw and the noise duration Ton, the period Tin for sampling data during normal processing and / or the number of continuous samplings for sampling data continuously for each sampling period Set Rn. Therefore, according to the result of analyzing the characteristics of the noise that is actually generated, the sampling cycle Tin and / or the number of continuous samplings Rn during normal processing can be set, and data sampling is performed while avoiding the influence of noise. Is possible.

  Further, the deviation value determination unit 15 of the control circuit 2 calculates an average value Tonave and a standard deviation value Tonσ for the noise duration Ton, and each value is less than a predetermined value (3 μsec, 1 μsec) determined for each. Sets the number of consecutive samplings Rn by the equation (2) based on the average value Tonave and the standard deviation value Tonσ. That is, in this case, since the noise ON width is concentrated at ± 3 μ of the average value Toave, it is possible to eliminate the influence of noise. In this case, when the number of consecutive samplings Rn becomes less than “3”, the calculation unit 7 sets the number of times Rn to “3”, so that the minimum number of data required for majority processing can be ensured. .

  In addition, when the average value Tonave and the standard deviation value Tonσ can exceed predetermined values determined for each, the calculation unit 7 obtains the average value Twave and the standard deviation value Twσ for the noise generation period Tw, and calculates the standard deviation When the value Twσ is less than a predetermined value and the average value Twave is less than the current sampling period Tin, the modulo operation of the expression (3) is performed, and the sampling period Tin is an average value based on the calculation result. If it is determined that it is close to an integral multiple of Twave, the sampling period Tin is changed. That is, in this case, there is a high possibility that sampling during normal processing is affected by noise that is periodically generated. Therefore, the possibility can be canceled by changing the sampling period Tin. In this case, the calculation unit 7 changes the sampling period Tin so that the value of the modulo calculation becomes “5”, so that the changed sampling period Tin does not become an integral multiple of the average value Twave of the noise generation period. It can be changed reliably.

  Further, when the standard deviation value Twσ is equal to or greater than the predetermined value, the calculation unit 7 obtains a frequency distribution of data values belonging to a class having a predetermined time width with respect to the noise generation period Tw, and calculates the sampling data number and the class number , The modulo operation is performed with the center value of the class as the average value Twave. That is, in this case, it can be reasonably assumed that the center value of the class is an average value of the noise generation period, so that sampling that reliably avoids the noise generation period can be performed based on the result of the modulo operation. It becomes possible.

  In addition, the arithmetic unit 7 sets the number of consecutive samplings Rn based on the average value Tonave and the standard deviation value Tonσ when there are a plurality of frequency classes determined to be significant. In this case, from the result of the frequency distribution, in order to reliably avoid the influence of noise, the sampling period Tin cannot be appropriately changed, so that the continuous sampling is performed from the average value Tonave and the standard deviation value Tonσ. The number of times Rn is set.

(Second Embodiment)
FIG. 17 shows an output example when the control circuit 2 outputs an alarm according to the second embodiment. The control circuit 2 uses the alarm output unit 16 to output an alarm as follows. When the alarm output unit 16 outputs 16-bit data as alarm data, for example, bit 0: Tin change alarm detection (execute noise data example A and S422)
Bit 1: Rn reliability alarm detection (Noise data example B, S430 executed)
Bit 2: Rn change alarm detection (Noise data example C, S412 is executed)
Bit 3: Rn response time exceeded alarm detection (S433 is executed)
And the alarm output is performed by setting the corresponding alarm bit to “1”. Thereby, the user can know what pattern of noise is generated.

  That is, when the noise cycle Tw is close to an integral multiple of the data sampling cycle Tin of the input terminal 4 as in the noise data example A, there is a high possibility that the input data value is erroneously recognized due to the influence of noise. . Further, when there are a plurality of noise periods as in the noise data example B, it is difficult to avoid the influence of noise by changing the sampling period Tin, so the control circuit 2 changes the number of consecutive samplings Rn. However, this is not necessarily a sufficient response. In the case of the noise data example C, the calculation result of the equation (1) is within a predetermined range, so that the noise period Tw is relatively close to the sampling period Tin and the standard deviation value Tonσ of the noise duration Ton and the average Since all the values Toave are less than the predetermined value, it is difficult to avoid the influence of noise by changing the sampling period Tin in this case as well. As a result, similar to the noise data example B, the number of continuous samplings Rn is changed to cope with this, but in this case, the possibility of being affected by noise is lower.

  As an actual alarm output, for example, display on the display means for displaying the frequency of the panel surface of the inverter device, output as an alarm signal to the output terminal prepared in the relay or open collector, On the other hand, it can be output by data communication. Accordingly, the data shown in FIG. 17 may be in either parallel / serial format.

  Further, when the user confirms the alarm output by communication or the like and the detection of the noise data example B is set, it is understood that the control circuit 2 is not performing the noise analysis satisfactorily. In this case, the frequency distribution data of the noise period Tw and the noise width Ton can be transmitted to the outside by communication or the like, and the noise can be analyzed by a host device such as a host computer. For example, it is determined whether or not the influence of noise is affected by whether or not the integer multiple of each noise period Tw is close to the sampling period Tin of the input terminal 4, and whether or not the sampling period Tin and the number of consecutive samplings Rn are changed. Or Tin and Rn which are not affected by noise can be determined. If the determined Tin and Rn are transmitted and set to the control circuit 2 by communication, the same effect can be obtained at a higher level.

  Similarly, the noise detection cycle Tw, the input data ON determination time, the input data OFF determination time, or their frequency distribution data can be read out to the outside. With such a configuration, an input signal terminal that is not affected by noise by calculating the sampling period Tin and the continuous sampling number Rn from the information read out to the outside by the host device and writing the result into the inverter device 1. Processing can be realized.

  As described above, according to the second embodiment, the calculation unit 7 sets the continuous sampling count Rn, changes the sampling cycle Tin (noise data example A), and changes the continuous sampling count Rn. About (noise data examples B and C), an alarm output is performed to the outside. In the latter case, different alarms occur when the noise generation period Tw is within a predetermined range including the sampling period Tin (noise data example B) and when there are a plurality of noise generation periods Tw (noise data example C). Was output. Therefore, the user can know what kind of noise is generated in the environment where the inverter device 1 is operating by referring to the alarm output result, and can take measures against noise. It becomes.

(Third embodiment)
18 and 19 show a third embodiment. In the inverter device 21 of the third embodiment, the output signal of the input circuit 5 is supplied to the arithmetic unit 23 that replaces the arithmetic unit 2 via the multiplexer 22, and the other input terminal of the multiplexer 22 includes Data output from the calculation unit 23 is given. The arithmetic unit 23 performs switching control of the multiplexer 22. The calculation unit 23 reads for noise diagnosis, stores it in the input data storage / save unit 8, outputs the data through the multiplexer 22 instead of the data input from the input terminal 4, and inputs the data to itself. After performing the noise diagnosis, the noise diagnosis is performed again to check whether or not to avoid erroneous detection.

FIG. 19 is an example of data that the arithmetic unit 23 stores in the input data storage / save unit 8 and outputs for verification. When the input process for turning ON the data of the input terminal 4 is performed when the data “1 (ON)” is recognized five times continuously from 0 μ to 4 μsec per hour, as shown in FIG. It can be confirmed that the data that should originally be OFF is recognized as ON at the timing when the recognition is “1”.
As described above, according to the third embodiment, the multiplexer 22 for inputting the data stored in the input data storage / save unit 8 is provided instead of the path for inputting the signal via the input terminal 4. Whether or not the influence of noise can be avoided can be verified based on the result of actually changing the sampling period Tin and the number of consecutive samplings Rn based on the noise information.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

For example, noise diagnosis or the like may be performed while the electric motor 6 is not stopped, where the processing load of the inverter device 1 is not so high.
The sampling period Tin, the number of continuous readings Rn, the majority number n for determining the data value, etc. may be appropriately changed. The same applies to the number of data samplings for analyzing noise information.
For example, the values used in step S460 and equation (1) do not necessarily need to be added with 5 times the standard deviation value, and may be changed as appropriate.
The noise data may be targeted for “0 (OFF)” when the default signal level is “1 (ON)”.

  In the drawings, 1 is an inverter device, 2 is a control circuit (sampling means, noise information detecting means, sampling setting means), 3 is an inverter main circuit, 4 is an input terminal, 6 is an electric motor, 7 is an arithmetic unit, and 8 is input data. A storage unit (storage unit), 16 an alarm output unit (alarm output unit), 22 a multiplexer (input switching unit), and 23 an arithmetic unit.

Claims (11)

In a control circuit that is mounted on an inverter device for variable speed control of an electric motor and performs control by reading a signal given to an input terminal,
Sampling means for sampling the binary level signal applied to the input terminal at regular intervals and storing the sampled data value in the storage means;
When the operation by the sampling means has elapsed for a predetermined time, the data value stored in the storage means is determined from the length of the period in which the data value “0” is continuous and the length of the period in which the data value “1” is continuous. Noise information detection means for detecting a noise generation period Tw estimated to be a change in the data value due to occurrence of the noise, and a noise duration Th in which the state in which the data value has changed continues,
Based on the noise generation period Tw and the noise duration time Th, the sampling means samples the data during normal processing and / or continuously at every sampling period so as to avoid the influence of noise. And a sampling setting means for setting the number of times Rn is sampled. The noise information detecting means obtains an average value Thave and a standard deviation value Thσ for the noise duration Th,
The sampling setting means sets the number of times Rn based on the average value Thave and the standard deviation value Thσ when the standard deviation value Thσ and the average value Thave are less than a predetermined value determined for each. The control circuit for an inverter device according to claim 1.   3. The control circuit for an inverter device according to claim 2, wherein the sampling setting means sets the number of times Rn to a value obtained by adding 5 times the standard deviation value Thσ to the average value Thave.   4. The control circuit for an inverter device according to claim 3, wherein the sampling setting means sets the number of times Rn to “3” when the number of times Rn becomes less than “3”. The sampling setting means obtains an average value Twave and a standard deviation value Twσ for the noise generation period Tw when the average value Thave and the standard deviation value Thσ can exceed predetermined values determined for each.
When the standard deviation value Twσ is less than a predetermined value and the average value Twave is less than the current sampling period Tin, the following modulo calculation is performed.
MOD (Tin * 10 / Twave, 10)
5. If the sampling cycle Tin is determined to be close to an integral multiple of the average value Twave based on the calculation result, the sampling cycle Tin is changed. 5. Control circuit.   6. The control circuit for an inverter device according to claim 5, wherein the sampling setting means changes the sampling period Tin so that a value of the modulo operation becomes “5”. When the standard deviation value Twσ is equal to or greater than the predetermined value, the sampling setting means obtains a frequency distribution of data values belonging to a class having a predetermined time width for the noise generation period Tw,
When there is only one frequency class that is determined to be significant from the number of sampling data and the class number, the modulo operation is performed with the center value of the class as the average value Twave. Item 7. An inverter device control circuit according to Item 5 or 6.   The sampling setting means sets the number of times Rn based on the average value Thave and the standard deviation value Thσ when there are a plurality of classes of frequency determined to be significant. Item 8. A control circuit for an inverter device according to Item 7.   9. The apparatus according to claim 5, further comprising an alarm output means for outputting an alarm to the outside when the sampling setting means sets the number of times Rn and when the sampling period Tin is changed. The control circuit of the inverter apparatus in any one.   In the case where the sampling setting unit sets the number of times Rn, the alarm output unit includes a case where the noise generation cycle Tw is within a predetermined range including the sampling cycle Tin and a plurality of the noise generation cycles Tw. The control circuit for an inverter device according to claim 9, wherein an alarm that is different from the case where the alarm is performed is output.   11. The input switching means for inputting data stored in the storage means to the sampling means instead of a path for inputting a signal through the input terminal. A control circuit for the inverter device according to claim 1.

Images