Program, information processing device, information processing method, and information processing system

Abstract

To provide a program that can accurately detect a sudden noise included in a target signal.SOLUTION: A program enables a computer to realize: an acquisition function for acquiring data on a plurality of channels, which is three or more; a selection function for selecting data on a single channel as attention channel data from the data on the plurality of channels acquired by the acquisition function; and a dissimilarity arithmetic function for computing dissimilarity of the attention channel data selected by the selection function to data on two or more channels among the data on the channels not selected by the selection function for each predetermined range of the attention channel data.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] This disclosure relates to a program, an information processing apparatus, an information processing method, and an information processing system.

Background Art

[0002] Measurement of electrical signals, magnetic signals, etc. using multiple sensors has been performed. Also, removal of noise included in the measurement signal has been performed.

[0003] In the multi-sensor signal abnormality detection device described in Patent Document 1, when measuring (detecting) a signal with multiple sensors, calculation of the noise level, which is a variation component between sensors, has been performed (see Patent Document 1).

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, in the conventional technology, there are cases where the accuracy of detecting and the like of the burst noise included in the measurement signal is insufficient. Here, burst noise is a sudden noise that occurs at a low frequency for each sensor. Burst noise may have an amplitude that is not extremely different from the peak of the signal of interest included in the measurement signal, for example.

[0006] In the conventional technology, when detecting and removing such burst noise, for example, the influence of burst noise is diluted by performing an additive average, or burst noise is discriminated by human visual inspection, and the location of the discriminated burst noise is excluded.

[0007] For example, in a measurement signal, peaks of the signal of interest may occur at predetermined intervals. In this case, if a sudden noise with a larger amplitude than the peak of the signal of interest occurs at the same time as the peak of the signal of interest, it becomes impossible to distinguish the peak of the signal of interest. On the other hand, when the amplitude of the sudden noise is small, it can be difficult to distinguish the sudden noise.

[0008] This disclosure has been made in consideration of these circumstances, and aims to provide a program, an information processing device, an information processing method, and an information processing system that can accurately detect sudden noise contained in a target signal. [Means for solving the problem]

[0009] One embodiment includes an acquisition function that acquires data from three or more channels, a selection function that selects data from a single channel as focus channel data from the data of the multiple channels acquired by the acquisition function, and a dissimilarity calculation function that calculates the dissimilarity of the focus channel data selected by the selection function with respect to data from two or more channels that were not selected by the selection function, for each predetermined range of the focus channel data. A composite data generation function that generates composite data which is the result of combining data from two or more channels that are not selected by the selection function, To make this possible on a computer A program wherein the dissimilarity calculation function calculates the dissimilarity to the composite data generated by the composite data generation function, and the composite data generation function generates the composite data by adjusting the weights for compositing data from two or more channels that have not been selected by the selection function, so that the composite data is close to the channel data of interest. It is a program.

[0010] One embodiment includes an acquisition unit that acquires data from three or more channels, a selection unit that selects data from a single channel as focus channel data from the data of the multiple channels acquired by the acquisition unit, and a dissimilarity calculation unit that calculates the dissimilarity of the focus channel data selected by the selection unit with respect to data from two or more channels that were not selected by the selection unit, for each predetermined range of the focus channel data. A composite data generation unit generates composite data which is the result of combining data from two or more channels that are not selected by the selection unit. Equipped with The dissimilarity calculation unit calculates the dissimilarity to the composite data generated by the composite data generation unit, and the composite data generation unit generates the composite data by adjusting the weights for combining data from two or more channels that have not been selected by the selection unit, so that the composite data is close to the target channel data. It is an information processing device.

[0011] In one embodiment, the acquisition unit of the information processing device acquires data from three or more channels, the selection unit of the information processing device selects data from a single channel from the data of the multiple channels acquired by the acquisition unit as the channel data of interest, and the dissimilarity calculation unit of the information processing device calculates the dissimilarity of the channel data of interest selected by the selection unit for each predetermined range of the channel data of interest, with respect to data from two or more channels that were not selected by the selection unit. The information processing device generates composite data, which is the result of combining data from two or more channels that have not been selected by the selection unit; the dissimilarity calculation unit calculates the dissimilarity of the composite data generated by the composite data generation unit; and the composite data generation unit generates the composite data by adjusting the weights for combining data from two or more channels that have not been selected by the selection unit so that the composite data is close to the target channel data. It is an information processing method.

[0012] One embodiment is an information processing system comprising a sensor unit that measures data from three or more channels, and an information processing device, wherein the information processing device includes an acquisition unit that acquires the data from the multiple channels measured by the sensor unit, a selection unit that selects the data from a single channel as the channel of interest from the data from the multiple channels acquired by the acquisition unit, and a dissimilarity calculation unit that calculates the dissimilarity of the data from two or more channels among the channel data not selected by the selection unit for each predetermined range of the channel of interest selected by the selection unit, A composite data generation unit generates composite data which is the result of combining data from two or more channels that are not selected by the selection unit. Equipped with The dissimilarity calculation unit calculates the dissimilarity to the composite data generated by the composite data generation unit, and the composite data generation unit generates the composite data by adjusting the weights for combining data from two or more channels that have not been selected by the selection unit, so that the composite data is close to the target channel data. It is an information processing system. [Effects of the Invention]

[0013] According to this disclosure, a program, an information processing device, an information processing method, and an information processing system can accurately detect sudden noise contained in a target signal. [Brief explanation of the drawing]

[0014] [Figure 1] This figure shows a schematic configuration of the information processing system according to the embodiment. [Figure 2] This figure shows an example of a functional block of the information processing apparatus according to the embodiment. [Figure 3] This figure shows an example of sudden noise. [Figure 4] This is a diagram showing an example of a measurement signal of the target channel P1 according to an embodiment. [Figure 5] This is a diagram showing an example of a measurement signal of the non-target channel P2 according to an embodiment. [Figure 6] This is a diagram showing an example of a measurement signal of the non-target channel P3 according to an embodiment. [Figure 7] This is a diagram showing an example of a measurement signal of the non-target channel P4 according to an embodiment. [Figure 8] This is a diagram showing an example of a composite signal according to an embodiment. [Figure 9] This is a diagram showing a comparison between the measurement signal of the target channel P1 and the composite signal according to an embodiment. [Figure 10] This is a diagram showing an example of a dissimilarity according to an embodiment. [Figure 11] This is a diagram showing an example of a method for determining a threshold value related to the dissimilarity according to an embodiment. [Figure 12] This is a diagram showing an example of spike noise included in the measurement signal according to an embodiment. [Figure 13] This is a diagram showing an example of block noise included in the measurement signal according to an embodiment. [Figure 14] This is a diagram showing an example of the influence of block noise when the measurement signal according to an embodiment is not averaged. [Figure 15] This is a diagram showing an example of suppression of the influence of block noise when the measurement signal according to an embodiment is averaged. [Figure 16] This is a diagram showing an example of a sensor unit having a grid-shaped multi-channel sensor according to a modification of an embodiment.

Embodiments for Carrying Out the Invention

[0015] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings.

[0016] [Information Processing System] FIG. 1 is a diagram showing a schematic configuration of an information processing system 1 according to an embodiment. For the sake of explanation, Figure 1 shows the XYZ coordinate system, which is a three-dimensional Cartesian coordinate system. In this embodiment, the direction parallel to the Z-axis is assumed to be parallel to the direction of gravity. Also, in this embodiment, the positive direction of the Z-axis is assumed to be upward, and the negative direction of the Z-axis is assumed to be downward.

[0017] The information processing system 1 comprises an information processing device 11, an A / D (Analog to Digital) converter 12, a display device 13, a belt conveyor 21, a sensor unit 31, a stand 41, and a current supply unit 42. The information processing system 1 may also be called, for example, a measurement system.

[0018] The sensor unit 31 is equipped with n sensors A-1 to An (where n is an integer greater than or equal to 3), which is three or more sensors. In this embodiment, each sensor A-1 to An measures the signal of its respective channel. A total of n channels of signals are measured by all sensors A-1 to An. Measurement may also be called testing or detection.

[0019] Multiple channels may also be referred to as multi-channel. Furthermore, the number of channels can be any value greater than or equal to three, for example, several dozen. Furthermore, in this embodiment, sudden noise may occur in the measurement signals of each sensor A-1 to An.

[0020] Furthermore, Figure 1 shows the object 51 to be measured. In this embodiment, the information processing system 1 does not include the object 51, but in other examples, the information processing system 1 may be considered to include the object 51. Figure 1 also shows the rotational axis 61 of the belt conveyor 21 and the rotational direction 62 of the belt conveyor 21.

[0021] The belt conveyor 21 is a device that rotates a wide, loop-shaped belt on a trolley (not shown) and moves objects placed on the belt. In the example shown in Figure 1, one face of the belt is positioned parallel to the XY plane. This face is the one on which the transported material is placed. In the example shown in Figure 1, the belt conveyor 21 has a circular shape with respect to a plane parallel to the XY plane and rotates in a predetermined rotational direction 62 around a rotational axis 61. In the example shown in Figure 1, the rotational axis 61 is parallel to the Z axis.

[0022] The sensor unit 31 has a linear multi-channel sensor. Such a multi-channel sensor may be called, for example, a sensor array. Each of the n sensors A-1 to An corresponds to a different channel. In the sensor unit 31, n sensors A-1 to An are arranged in a straight line at predetermined intervals. These predetermined intervals may be, for example, equal intervals. When viewed from a direction parallel to the Z-axis, this straight line intersects with a portion of the belt of the conveyor belt 21. At this intersection, the straight line is perpendicular and orthogonal to the rotational direction of the portion of the belt. In the example in Figure 1, the m-th sensor Am among the n sensors A-1 to An is positioned near the center of the belt in the radial direction. Here, if n is odd, m is the number of the center sensor, i.e., (n+1) / 2, and if n is even, m is a number near the center sensor, i.e., n / 2 or (1+n / 2).

[0023] Each sensor A-1 to An measures the same physical quantity. This physical quantity may be any quantity, such as an electrical signal, a magnetic signal, or an audio signal. In such n-channel data, for example, there may be similarities between two adjacent channels, or between two or more adjacent channels.

[0024] In this embodiment, each of sensors A-1 to An is a sensor (magnetic sensor) that measures magnetic signals. In this embodiment, each of sensors A-1 to An is an example of a foreign object detection sensor that detects foreign objects. Each sensor A-1 to An outputs a measurement result signal (measurement signal) in analog signal format to the A / D converter 12. The measurement signal may be, for example, a voltage signal. Furthermore, among the multiple sensors A-1 to An, there may be sensors that measure different physical quantities.

[0025] Platform 41 is placed on a designated location on the upper surface of the belt conveyor 21. Furthermore, a current supply unit 42 is provided on the top surface of the base 41. Furthermore, the object 51 is placed on the top surface of the stand 41. In this embodiment, the relative position of the object 51 with respect to the belt conveyor 21 remains unchanged even when the belt conveyor 21 rotates. Furthermore, in this embodiment, the relative position of the platform 41 with respect to the belt conveyor 21 also remains unchanged.

[0026] In the example shown in Figure 1, as the belt conveyor 21 rotates, the object 51 also rotates in a predetermined direction 62 around the rotation axis 61. As a result, from a viewpoint parallel to the Z-axis, the object 51 rotates along the circular path of the belt conveyor 21. In this embodiment, the rotational speed of the belt conveyor 21 is assumed to be constant. However, as an example of another configuration, a configuration in which the rotational speed of the belt conveyor 21 can change may be used. In this case, the information processing device 11 or the like may perform processing that takes into account the change in rotational speed (for example, adjusting the rate at which the time axis values ​​progress).

[0027] Here, object 51 is the object to be measured, for example, a sample that is subject to inspection in a factory or similar facility. In this embodiment, the current supply unit 42 continuously supplies current to the object 51. This current may be, for example, a current of a constant magnitude. As a result, the object 51 continuously generates a magnetic signal. This magnetic signal is then measured by multiple sensors A-1 to An.

[0028] In this embodiment, if foreign matter or damage is present on the object 51, the object 51 generates a magnetic field different from the magnetic field that is generated under normal conditions. Here, "under normal conditions" refers to the case where no foreign matter or damage is present on the object 51. The object 51 may be any sample that conducts electric current, for example, a metal material part. On the other hand, if there are no foreign objects or scratches on the object 51, the object 51 generates magnetism that occurs under normal conditions. The waveform of the magnetic signal measured by sensors A-1 to An due to this magnetism may be acquired in advance, for example, by predetermined calculations.

[0029] In the example shown in Figure 1, the object 51 is rotated and, upon reaching a location where multiple sensors A-1 to An are present, passes across the line in which these sensors A-1 to An are arranged. At this time, each of the sensors A-1 to An measures the magnetic signal generated by the magnetism (magnetic field) produced by the object 51.

[0030] Here, each sensor A-1 to An can measure a magnetic signal even when the object 51 is at a distance, but in this embodiment, it is assumed that a particularly large peak in the magnetic signal is measured when the object 51 is at a close distance. In this embodiment, each time the object 51 passes across a line where multiple sensors A-1 to An are arranged, one particularly large magnetic signal peak is measured.

[0031] Furthermore, in this embodiment, the belt conveyor 21 rotates multiple times for the same object 51. As a result, in the magnetic signals measured by each sensor A-1 to An, if no sudden noise occurs, a similar peak will appear each time the belt conveyor 21 rotates for one full turn. Such peaks will appear periodically if the rotation speed of the belt conveyor 21 is constant. However, this embodiment shows a case in which sudden noise may occur in the magnetic signals measured by each of the sensors A-1 to An.

[0032] As another example, in this embodiment, the magnetic signal of the same object 51 may be processed when the object 51 passes across a line of multiple sensors A-1 to An once. In this case, for example, one peak in the magnetic signal may be measured by each sensor A-1 to An, or multiple peaks in the magnetic signal may be measured by each sensor A-1 to An, and then the information processing device 11 may extract and process the magnetic signal portion of one peak from these measurement results.

[0033] In the example shown in Figure 1, the current supply unit 42 is mounted on the base 41, but as an example of other configurations, the current supply unit 42 may be mounted at any other location. Furthermore, while this embodiment shows a case where the current supply unit 42 constantly supplies current to the object 51, another configuration example is one in which the current supply unit 42 supplies current to the object 51 when the object 51 is near the sensor unit 31. In other words, it is sufficient that the magnetic signal generated by the object 51 can be measured by multiple sensors A-1 to An.

[0034] Furthermore, while the example in Figure 1 shows a platform 41 placed on the top surface of the belt conveyor 21 and an object 51 placed on the top surface of the platform 41, other configurations are also possible, where the object 51 is placed on the top surface of the belt conveyor 21.

[0035] The A / D converter 12 receives the analog signals output from each of the sensors A-1 to An. The A / D converter 12 converts the analog signals input from each sensor A-1 to An into digital signals using A / D conversion, and inputs the resulting digital signals to the information processing device 11.

[0036] In this embodiment, an A / D converter 12 is provided between the sensor unit 31 and the information processing device 11, but other configurations may be used. For example, the A / D conversion function of the A / D converter 12 may be provided in the sensor unit 31, and the measurement result signals from each sensor A-1 to An may be input to the information processing device 11 in the form of digital signals. As another example, the A / D conversion function of the A / D converter 12 may be provided in the information processing device 11, and the measurement result signals from each sensor A-1 to An may be input to the information processing device 11 in the form of analog signals, and these analog signals may be converted into digital signals by the A / D conversion function.

[0037] The information processing device 11 receives the digital signal (data) output from the A / D converter 12. The information processing device 11 performs predetermined processing using the input digital signal (data). Furthermore, the information processing device 11 outputs the information to be displayed to the display device 13.

[0038] The display device 13 has a screen and a function to display information on the screen. In this embodiment, the display device 13 receives information output from the information processing device 11 and displays the information on the screen. As another example, the display device 13 may print the information output from the information processing device 11, thereby displaying the information as a printed document on paper or the like.

[0039] In this embodiment, for the sake of simplicity, we show a case where one object 51 is placed on the belt conveyor 21. However, in other configuration examples, objects may be placed on different parts of the belt conveyor 21, resulting in a total of multiple objects being placed on it. In this case, for example, each object may be processed in the same way as the object 51 in this embodiment. Furthermore, in this embodiment, for the sake of simplicity, the case in which the sensor unit 31 is provided at one location on the belt conveyor 21 is shown. However, as an example of other configurations, the sensor unit may be provided at each of multiple locations on the belt conveyor 21. In this case, for example, the same processing as that performed on the sensor unit 31 in this embodiment may be performed on each sensor unit.

[0040] <Information Processing Device> Figure 2 shows an example of a functional block of the information processing device 11 according to the embodiment. The information processing device 11 includes an input unit 111, an output unit 112, a communication unit 113, a storage unit 114, and a processing unit 115. The processing unit 115 includes an acquisition unit 131, a selection unit 132, a composite data generation unit 133, a dissimilarity calculation unit 134, a dissimilarity determination unit 135, a defective range determination unit 136, a defective range removal unit 137, a defective data exclusion unit 138, and a threshold determination unit 139.

[0041] In this embodiment, the information processing device 11 is configured using a computer that operates using hardware such as a processor and software such as a program. The information processing device 11 is equipped with a processor such as a CPU (Central Processing Unit), and performs various processes by executing a predetermined program using this processor. The program may be stored, for example, in the memory unit 114.

[0042] The input unit 111 is used to input information. The input unit 111 may, for example, include an operation unit such as a keyboard and a mouse, and input information corresponding to the operations performed by the user on the operation unit. Furthermore, the input unit 111 may be connected to an external device, for example, and receive information output from that device. This device may be, for example, a portable storage device.

[0043] The output unit 112 outputs information. The output unit 112 may be connected to an external device, for example, and output information to that device. This device may be, for example, a display device 13. Alternatively, this device may be, for example, a portable storage device.

[0044] The communication unit 113 has the function of performing communications. In this embodiment, the receiving function by the communication unit 113 and the input unit 111 are shown as separate functional units. However, in other configuration examples, the receiving function by the communication unit 113 may be included in the input unit 111. Similarly, this embodiment shows a case where the transmission function by the communication unit 113 and the output unit 112 are provided as separate functional units, but as another example of configuration, the transmission function by the communication unit 113 may be included in the output unit 112.

[0045] The memory unit 114 stores information. In this embodiment, the information processing device 11 is shown to use an internal storage unit 114, but as another example of configuration, the information processing device 11 may use an external storage unit (not shown).

[0046] The acquisition unit 131 acquires data (measurement result data) from three or more channels. In this embodiment, the acquisition unit 131 acquires the digital signal (measurement result data) input from the A / D converter 12 to the information processing device 11.

[0047] In this embodiment, the information processing device 11 receives the digital signal (measurement result data) output from the A / D converter 12 via the input unit 111 (or the receiving function of the communication unit 113).

[0048] The selection unit 132 selects data from a single channel (i.e., one channel) from the data of multiple channels acquired by the acquisition unit 131 as the channel data of interest. In this embodiment, the selection unit 132, for example, selects data from each of the n channels one by one in a predetermined order, thereby selecting all the data from all channels in total. The channels of interest may also be called target channels.

[0049] The composite data generation unit 133 generates composite data, which is the result of combining data from two or more channels that were not selected by the selection unit 132. Here, the composite data generation unit 133 may, for example, generate composite data which is the result of combining all the data of channels that have not been selected by the selection unit 132. Furthermore, the composite data generation unit 133 may generate composite data that is the result of combining all of the normal data (for example, data that has not been determined to be defective) from the data of channels that have not been selected by the selection unit 132.

[0050] The composite data generation unit 133 may generate composite data by adjusting the weights for compositing data from two or more channels that were not selected by the selection unit 132, for example, by least-squares error approximation on the channel data of interest. In other words, when the composite data generation unit 133 synthesizes data from two or more channels that have not been selected by the selection unit 132, it may adjust the weight of each data to be synthesized so that the synthesized data is close to (for example, the closest to) the data of the channel of interest.

[0051] In the least squares error approximation, for the two data sets being compared (in this example, the channel data of interest and the composite data), the sum of the squared differences between corresponding values ​​(values ​​that form pairs between the two data sets) is calculated for all pairs (sum of squared errors). The weights used when creating the composite data are then adjusted so that this sum of squared errors is minimized. In this example, the corresponding values ​​are those from the same sample (same time). In this embodiment, the least squares error approximation method is used to generate data that is close to the channel data of interest as composite data, but other approximation methods may be used.

[0052] The dissimilarity calculation unit 134 calculates the dissimilarity of the data from two or more channels that were not selected by the selection unit 132 for each predetermined range of the channel data selected by the selection unit 132. Here, the dissimilarity calculation unit 134 may calculate the dissimilarity to the composite data generated by the composite data generation unit 133 as the dissimilarity. In this embodiment, the dissimilarity is a value used to evaluate the degree to which two data points are not similar. Dissimilarity may also be called, for example, deviation or distance.

[0053] Alternatively, instead of dissimilarity, similarity may be used, which is a value used to evaluate the degree to which two data points are similar. In this case, for example, a higher similarity score may be considered to indicate a lower dissimilarity score, and a lower similarity score may be considered to indicate a higher dissimilarity score.

[0054] The specified range may be two or more different ranges. The predetermined range may be the range of one sample of the channel data of interest.

[0055] The data from multiple channels may each be time-series data. In this case, the specified range may be a range of time periods. Another possible configuration is that the data for multiple channels may be data arranged according to their spatial configuration. In this case, the predetermined range may be a spatial range.

[0056] Various methods may be used for calculating the dissimilarity using the dissimilarity calculation unit 134. As an example, the dissimilarity calculation unit 134 may calculate the dissimilarity based on the sum of the absolute differences between the channel data of interest and the composite data. As an example, the dissimilarity calculation unit 134 may calculate the dissimilarity based on the sum of the squared differences between the channel data of interest and the composite data. As an example, the dissimilarity calculation unit 134 may calculate the dissimilarity based on the maximum absolute value of the difference between the channel data of interest and the composite data. As an example, the dissimilarity calculation unit 134 may calculate dissimilarity based on the difference between the features of the channel data of interest and the features of the composite data. As an example, the dissimilarity calculation unit 134 may calculate the dissimilarity based on the difference between the probability distribution of the channel data of interest and the probability distribution of the composite data. As an example, the dissimilarity calculation unit 134 may calculate the dissimilarity based on the difference between the power spectrum of the channel data of interest and the power spectrum of the composite data. As an example, the dissimilarity calculation unit 134 may calculate the dissimilarity using a calculation formula that associates samples from the channel data of interest with the composite data.

[0057] The dissimilarity determination unit 135 determines whether the dissimilarity calculated by the dissimilarity calculation unit 134 is higher than a predetermined threshold.

[0058] The defective range determination unit 136 determines that the range determined by the dissimilarity determination unit 135 to have a dissimilarity higher than a threshold is a defective range. In this embodiment, the case where the dissimilarity determination unit 135 and the defect range determination unit 136 are separate functional units is shown. However, as another example of configuration, the dissimilarity determination unit 135 and the defect range determination unit 136 may be configured as the same functional unit (common functional unit).

[0059] The defective range removal unit 137 replaces the data portion of the range determined as a defective range by the defective range determination unit 136 with the data portion resulting from the synthesis of data from two or more channels among the channels not selected by the selection unit 132.

[0060] When the defective data exclusion unit 138 performs a predetermined process based on multiple instances of the selected channel data, it excludes the selected channel data that includes the range determined to be a defective range by the defective range determination unit 136. Here, various processes may be used as the predetermined process; for example, an averaging process may be used.

[0061] Furthermore, the processing of the defective range removal unit 137 and the processing of the defective data exclusion unit 138 may be either one of the two. In this case, the functional unit for the other processing (defective range removal unit 137 or defective data exclusion unit 138) does not need to be provided in the information processing device 11.

[0062] The threshold determination unit 139 has the function of determining a threshold based on the statistical results of the dissimilarity of each range of the channel data of interest. Here, the predetermined threshold used by the dissimilarity determination unit 135 may be, for example, set in advance, or determined by the threshold determination unit 139.

[0063] The information processing device 11 may output information for displaying measurement result data acquired by the acquisition unit 131 of the processing unit 115, data during processing performed by the processing unit 115, data resulting from processing performed by the processing unit 115, or data related thereto, to the display device 13 via the output unit 112, thereby displaying the information on the display device 13.

[0064] In this embodiment, the display device 13 and the information processing device 11 are shown as separate components. However, as another example of a configuration, the display device 13 may be integrated with the information processing device 11.

[0065] [Sudden noise] Figure 3 shows an example of sudden noise. In the graph shown in Figure 3, the horizontal axis represents time, and the vertical axis represents level. This level is, for example, the magnitude of the signal and can take on both positive and negative values. Figure 3 shows an example of a measurement signal 1011 from one channel. Figure 3 also shows an example of a sudden noise 2011 that occurred in the measurement signal 1011 of the channel in question.

[0066] The transient noise 2011 can take on either a positive or negative value. In the example in Figure 3, the transient noise 2011 takes on a negative value. Furthermore, in the example shown in Figure 3, the measurement signal 1011 drops to the negative side due to the influence of the sudden noise 2011, but afterwards, it returns to a state where the influence of the sudden noise 2011 is absent (this is a state that has progressed further than shown in Figure 3, and is therefore omitted from the illustration). Note that the measurement signal 1011 and the sudden noise 2011 shown in Figure 3 are illustrative examples for illustrative purposes and do not necessarily represent the exact signal waveforms.

[0067] As a specific example, in MR sensors, which are magnetic sensors that utilize the magnetoresistance (MR) effect, sudden noise can occur due to jumps in the internal state of the sensor module. Such sudden noise may be called, for example, jump-like noise. In MR sensors, a small magnet is placed inside the sensor element, and sudden noise occurs when the magnetic field of this magnet reverses.

[0068] The sudden noise generated at the output of each channel is unique to that channel and does not affect the output of other channels. Sudden noise occurs when the output level changes abruptly, like a jump. The frequency of sudden noise occurrences is low. A specific example of the frequency of sudden noise occurrences is less than once every 100 cycles of the signal of interest (the signal of interest), but this is an example for illustrative purposes only and is not limited to this. Sudden noise occurs abruptly, not periodically.

[0069] [Specific examples of processing measurement signals] Referring to Figures 4 to 10, a specific example of the processing of measurement signals performed by the information processing device 11 is shown. Note that the signal waveforms, such as the measurement signals shown in Figures 4 to 10, are illustrative examples for explanatory purposes and do not necessarily represent the exact signal waveforms. In this embodiment, the measurement signal becomes the signal (target signal) that is to be detected (determined) as sudden noise.

[0070] <Measurement signals from 4 channels> In this example, to simplify the explanation, we will describe the processing of measurement signals from four channels as an example of processing measurement signals from multiple channels. In this example, the four channels will be referred to as channel P1, channel P2, channel P3, and channel P4. In this example, we will refer to the one channel that is attracting attention among these four channels P1 to P4 as the "attention channel," and the other three channels (the channels that are not attracting attention) as "non-attention channels." The attention channel can be switched among the four channels P1 to P4.

[0071] Figures 4 to 10 illustrate the case where channel P1 is the channel of interest. In this example, Channel P1, Channel P2, Channel P3, and Channel P4 will be referred to as the focus channel P1, the non-focus channel P2, the non-focus channel P3, and the non-focus channel P4, respectively. Furthermore, this example shows a case where sudden noise occurs in the measurement signal of one channel P1, but not in the measurement signals of the other three channels.

[0072] The information processing device 11 acquires data of measurement signals from four channels P1 to P4 using the acquisition unit 131. In this example, the information processing device 11, using the selection unit 132, selects the data of the measurement signal of one channel P1 as the data of the measurement signal of the channel P1 of interest.

[0073] Figure 4 shows an example of the measurement signal 1021 of the channel P1 of interest according to the embodiment. Figure 5 shows an example of the measurement signal 1031 of the non-focus channel P2 according to the embodiment. Figure 6 shows an example of the measurement signal 1041 of the non-focus channel P3 according to the embodiment. Figure 7 shows an example of the measurement signal 1051 of the non-focus channel P4 according to the embodiment. In the graphs shown in Figures 4 to 7, the horizontal axis represents time, and the vertical axis represents level. This level is, for example, the magnitude of the signal and can take on both positive and negative values.

[0074] As shown in Figure 4, the measurement signal 1021 of the channel P1 of interest exhibits sudden noise. Specifically, the peak positions 3011-3015 that appear periodically (or nearly periodically) in the measurement signal 1021 are the peak positions of the signal of interest included in the measurement signal 1021. Furthermore, the measurement signal 1021 includes transient noise 2021 and transient noise 2022.

[0075] In the measurement signal 1031 of the channel P2 shown in Figure 5, no sudden noise occurs. Specifically, the peak positions 3031 to 3035 that appear periodically (or nearly periodically) in the measurement signal 1031 are the peak positions of the signal of interest included in the measurement signal 1031.

[0076] In the measurement signal 1041 of the channel P3 shown in Figure 6, no sudden noise is present. Specifically, the peak positions 3041-3045 that appear periodically (or nearly periodically) in the measurement signal 1041 are the peak positions of the signal of interest included in the measurement signal 1041.

[0077] In the measurement signal 1051 of the channel P4 shown in Figure 7, no sudden noise is present. Specifically, the peak positions 3051-3055 that appear periodically (or nearly periodically) in the measurement signal 1051 are the peak positions of the signal of interest included in the measurement signal 1051.

[0078] In the example shown in Figure 1, the peaks of the interest signal included in the measurement signals 1021, 1031, 1041, and 1051 of each channel P1 to P4 occur when the object 51 is near the sensor unit 31. Each time the object 51 passes the sensor unit 31, one peak of the interest signal is generated. In the examples shown in Figures 4 to 7, the object 51 passes over the sensor unit 31 five times, and five peaks of the signal of interest occur periodically (or nearly periodically).

[0079] In this embodiment, the signals of interest measured by multiple channels P1 to P4 are due to magnetism generated from the same object 51, and are signals measured by each sensor at a different position. Therefore, although the signals of interest measured on multiple channels P1 to P4 do not have the same waveform, the timing (temporal position) at which the peaks of the signals of interest appear on multiple channels P1 to P4 is the same or close. Also, the period at which the peaks of the signals of interest appear on multiple channels P1 to P4 is the same or close. Note that the peaks of the signal of interest measured across multiple channels P1-P4 may include peaks with opposite positive and negative directions.

[0080] <Generating Synthetic Data> The information processing device 11 generates composite data, which is the result of combining the data of the measurement signals (measurement results) of the three non-focus channels P2 to P4, using the composite data generation unit 133. In this example, the composite data generation unit 133 adjusts the weights of the three non-focus channels P2 to P4 by applying the least-squares error approximation to the data of the measurement signal of the channel of focus P1, and generates composite data which is the result of combining the measurement signal data of the three non-focus channels P2 to P4. These weights contribute to the synthesis (addition). For example, if the weights of non-focused channels P2 to P4 are 0.5, 0.3, and 0.2 respectively, the result of multiplying the data of non-focused channel P2 by 0.5, the result of multiplying the data of non-focused channel P3 by 0.3, and the result of multiplying the data of non-focused channel P4 by 0.2 will be synthesized.

[0081] In this example, a linear combination is used for synthesis, but other synthesis methods may also be used. Furthermore, this example demonstrates a method where the synthesized data is made to most closely resemble the data of the target channel P1, but other synthesis methods may also be used. Furthermore, in this example, the data for the measurement signal of channel P1 of interest and the measurement signal data of channels P2 to P4 of non-interest of interest are combined using the entire time range of the measurement signals. However, other ranges may be used for the combination. Note that the time range used for the combination is common to all channels P1 to P4, for example.

[0082] In this example, we have shown a case where the weights of the synthesis are adjusted, but as an example of other configurations, if there is no practical problem, a predetermined fixed weight may be used, or a configuration in which the weights of all channels are the same value (a configuration in which weights are not used in effect) may be used. However, in this embodiment, since sudden noise can occur on both the positive and negative sides, weighting is necessary. Here, the weight values ​​do not necessarily have to be positive; negative values ​​may be used, or both positive and negative values ​​may be used.

[0083] Figure 8 shows an example of the synthesized signal 1111 according to the embodiment. Here, the composite signal 1111 is a signal representing the composite data generated by the composite data generation unit 133. In the example in Figure 8, the composite signal 1111 shows five peak positions 3111 to 3115.

[0084] <Comparison of measured signal and composite signal of the channel of interest> The dissimilarity calculation unit 134 calculates the dissimilarity of the measurement signal data of the channel of interest P1 for each predetermined range, relative to the measurement signal data of the unselected non-interest channels P2 to P4. In this example, the dissimilarity calculation unit 134 calculates the dissimilarity relative to the composite data. In this example, the predetermined range is defined as a time period with a predetermined duration. The frame representing this predetermined range may be called a window or the like.

[0085] Figure 9 shows a comparison between the measurement signal 1021 and the composite signal 1111 of the channel P1 of interest according to this embodiment. Here, the measurement signal 1021 is the same as the measurement signal 1021 shown in Figure 4. Furthermore, the combined signal 1111 is the same as the combined signal 1111 shown in Figure 8.

[0086] In the example shown in Figure 9, the time axis (horizontal axis) of the measured signal 1021 and the time axis (horizontal axis) of the composite signal 1111 are aligned. Figure 9 shows a window 313 representing the range for calculating dissimilarity. Window 313 is the range between the lower limit 311 and the upper limit 312 on the time axis (horizontal axis). In this example, the width of this range is constant.

[0087] Initially, the dissimilarity calculation unit 134 sets a window (not shown) with a certain width, using the minimum value of the time axis in which the measurement signal 1021 and the composite signal 1111 exist as the lower limit, and calculates the dissimilarity between the signal portion of the measurement signal 1021 and the signal portion of the composite signal 1111 included in the window. Subsequently, the dissimilarity calculation unit 134 keeps the width of the window constant and moves the window in the movement direction 361 (movement directions 362 and 363 also represent the same direction) for a predetermined time interval, and calculates the dissimilarity between the signal portion of the measurement signal 1021 contained in the window and the signal portion of the composite signal 1111 contained in the window.

[0088] Subsequently, the dissimilarity calculation unit 134 repeats the window movement and dissimilarity calculation in the same manner. The dissimilarity calculation unit 134 then repeatedly moves the window and performs dissimilarity calculations until the upper limit of the window exceeds the maximum value of the time axis in which the measurement signal 1021 and the composite signal 1111 exist.

[0089] Here, any time interval (time width) can be used as the predetermined time interval for moving the window. For example, a time interval the same size as the width of the window may be used, or a time interval that is half the width of the window (= 1 / 2) may be used. Furthermore, the predetermined time interval may not only be a time interval for multiple samples, but also a time interval for one sample. For example, in cases where the time width of the sudden noise is assumed to be (1 / 1000) seconds or less, an embodiment in which the time width for one sample is (1 / 1000) seconds may be used. In this embodiment, the time interval for one sample corresponds to the time interval for one sample of the measurement signal 1021, and the time interval for one sample of the composite signal 1111 is the same.

[0090] Adjacent windows do not necessarily have to overlap, or they may partially overlap. Adjacent windows also do not need to be separated from each other. For example, if the time interval between moving a window is the same as the width of the window, then adjacent windows will touch each other. For example, if the time interval between moving a window is smaller than the width of the window, adjacent windows will partially overlap. For example, if the time interval between moving windows is greater than the width of the window, adjacent windows will move apart from each other.

[0091] For example, in a configuration where the time interval for moving the windows is half the width of the window, it is possible to suppress the oversight of sudden noise while keeping the total number of windows low.

[0092] In the example in Figure 9, window 323 corresponds to the window into which window 313 has been moved. Window 323 represents the range between the lower limit 321 and the upper limit 322 on the time axis (horizontal axis). Note that windows 313 and 323 are examples of the positions of two windows during the movement process, and windows at other positions are not shown in the illustration. A window that can be moved in this manner may be called, for example, a sliding window. The time interval for moving the window may be called, for example, a step width.

[0093] Furthermore, in the example shown in Figure 9, the comparison between the signal portions contained in window 313 of the measurement signal 1021 and the composite signal 1111 is schematically shown by the arrow of comparison 351. Similarly, in the example in Figure 9, the comparison between the signal portions contained in window 323 of the measurement signal 1021 and the composite signal 1111 is schematically shown by the arrow of comparison 352.

[0094] In this example, we have shown the case where two or more different windows are used. However, as an alternative configuration, the width of the window may be the width of the entire time period in which the measurement signal 1021 exists (in this example, the width of the entire time period in which the composite signal 1111 exists is also the same). In this case, the window is set only once and does not move.

[0095] <Methods for calculating dissimilarity> The dissimilarity calculation unit 134 calculates the dissimilarity for each window. Here, any calculation method may be used for calculating dissimilarity.

[0096] As an example, the sum of the absolute differences between the channel data of interest and the composite data may be used as the dissimilarity calculation method. This sum is the sum of the absolute differences between corresponding values ​​in these two sets of data. In this example, the corresponding values ​​are those from the same sample (same time).

[0097] As an example, the sum of the squares of the differences between the channel data of interest and the composite data may be used as the dissimilarity calculation method. This sum of squares is the sum of the squares of the differences between corresponding values ​​in these two sets of data. In this example, the corresponding values ​​are those from the same sample (same time).

[0098] As an example, the maximum absolute value of the difference between the channel data of interest and the composite data may be used as the dissimilarity calculation method. In this example, the maximum value is determined from the values ​​for each sample (each time) within the window.

[0099] As an example, the difference between the features of the channel data of interest and the features of the composite data may be used as the dissimilarity calculation method. Here, the feature quantities for each data point may be calculated, for example, based on the waveform representing each of those data points. Various features may be used as features, for example, the mean, variance, various higher-order statistics, or a combination of two or more of these statistics.

[0100] As an example, a value representing the difference (distance) between the probability distribution of the channel data of interest and the probability distribution of the composite data may be used as the dissimilarity calculation method. Here, the probability distribution of each data point is, for example, the probability distribution followed by the sample values ​​contained in the waveform representing each of those data points. Various probability distributions may be used as the probability distribution, such as KL-divergence, JS-divergence, Pearson distance, relative Pearson distance, or L2 distance.

[0101] As an example, the difference between the power spectrum of the channel data of interest and the power spectrum of the composite data may be used as the dissimilarity calculation method. Here, the power spectrum of each data point is, for example, the power spectrum of the waveform representing that particular data point. Furthermore, the difference in the power spectra of two data points may be, for example, the difference in power at a specific frequency or a specific frequency band. In this case, the specific frequency may include multiple different frequencies, and the specific frequency band may include multiple different frequency bands.

[0102] As an example, a dissimilarity calculation method may use a value obtained by a calculation formula that associates samples from the channel data of interest with the composite data (the value calculated by said formula) as the dissimilarity. Here, such dissimilarity may be a value based on, for example, the value obtained by a cross-correlation function. Note that the value obtained by the cross-correlation function is the similarity, so for example, the dissimilarity value may decrease as the cross-correlation function value increases.

[0103] <Dissimilarity> Figure 10 shows an example of dissimilarity according to the embodiment. In the graph shown in Figure 10, the horizontal axis represents time, and the vertical axis represents dissimilarity. In this embodiment, a higher dissimilarity value indicates a greater degree of dissimilarity.

[0104] Figure 10 shows the dissimilarity characteristic 1211, which is the characteristic of dissimilarity with respect to time, and the dissimilarity threshold 411. Here, the dissimilarity characteristic 1211 is a characteristic obtained by connecting the dissimilarity calculated over time along the time axis. Furthermore, the threshold 411 may be, for example, pre-set, or it may be a value determined by the threshold determination unit 139.

[0105] In the example in Figure 10, the dissimilarity represented by the dissimilarity characteristic 1211 exceeds the threshold 411 at two points in time (which have a time interval in the example in Figure 10). In the example shown in Figure 10, these two locations are schematically represented as the sudden noise section 431 and the sudden noise section 432. The dissimilarity determination unit 135 determines whether the dissimilarity calculated by the dissimilarity calculation unit 134 is higher than the threshold 411. In the example in Figure 10, the dissimilarity determination unit 135 determines that the dissimilarity is higher than the threshold 411 at each time interval of the sudden noise section 431 and the sudden noise section 432 (in the example in Figure 10, these intervals have a width).

[0106] <Method for determining the threshold for dissimilarity> Figure 11 shows an example of a method for determining a threshold for dissimilarity according to the embodiment. Referring to Figure 11, an example of a method for determining a threshold for dissimilarity (threshold 411 in the example of Figure 10) by the threshold determination unit 139 is shown.

[0107] In the graph shown in Figure 11, the horizontal axis represents dissimilarity, and the vertical axis represents frequency. In other words, the graph represents a histogram of dissimilarity. In the example in Figure 11, there is a region where a large number of frequencies are concentrated on the side with low dissimilarity. In this example, this region is referred to as the normal noise region 511 and will be explained accordingly. In this example, normal noise represents noise other than sudden noise, for example, noise that is not as sudden as sudden noise.

[0108] In the example in Figure 11, there are parts with very few frequencies in the high-dissimilarity region. In this example, we will refer to this region as the sudden noise region 512, which is the sudden noise region, and explain it accordingly. The threshold determination unit 139 then determines the value between the normal noise range 511 and the sudden noise range 512 as the threshold 521.

[0109] In general terms, the threshold determination unit 139 determines the threshold value 521 as the result of adding a predetermined margin to the upper limit of dissimilarity when no sudden noise is present. For example, the upper limit of dissimilarity in the normal noise range 511 may be used as this upper limit. Here, there are no particular limitations on the range of dissimilarity in which the noise range 511 is normally set; for example, a range in which the frequency is greater than or equal to a predetermined value may be used. Furthermore, there are no particular limitations on the range of dissimilarity in which the sudden noise range 512 is set; for example, a range in which the frequency is less than a predetermined value may be used.

[0110] As an example, the threshold determination unit 139 may determine the threshold 521 as a value 3.00 to 3.05 times the dissimilarity of the centroid (or mean, etc.) of the normal noise range 511. Based on empirical rules, such a value is a good threshold for detecting sudden noise. As another example, the threshold determination unit 139 may determine the threshold 521 using various distribution functions or variances. For example, outliers based on variances may be used as the threshold 521. In this way, the threshold determination unit 139 determines the threshold 521 based on the statistical results of the dissimilarity of each range (in this example, the time range) of the data of the channel of interest P1.

[0111] In this example, for the sake of simplicity, the window range used to determine the defective range and the window range used to determine the threshold in the measurement signal data are the same; however, different ranges may be used.

[0112] <Processing related to the scope of defects> The defective range determination unit 136 determines that the range in which the dissimilarity is determined to be higher than the threshold is a defective range. In this embodiment, the information processing device 11 performs, for example, the processing of the defective range unit 137 or the processing of the defective data exclusion unit 138 as processing related to the defective range.

[0113] <Processing to remove defective areas> For the sake of explanation, the process of removing defective areas will be explained with reference to Figure 9. In this example, we will explain the case where the area of ​​window 323 shown in Figure 9 is determined to be a defective area. The defective range removal unit 137 replaces the data portion of the measurement signal 1021 of the channel P1 of interest that is determined to be a defective range with the data portion of the composite signal 1111 that is the same range as the defective range.

[0114] In the example shown in Figure 9, the defective area removal unit 137 replaces the data portion of the measurement signal 1021 within the range of window 323 where sudden noise is considered to have occurred with the data portion of the composite signal 1111 within the range of window 323. As a result, the data portion where sudden noise is considered to have occurred is removed from the data of the measurement signal 1021 of the channel of interest P1.

[0115] In this example, for the sake of simplicity, the range of the window used to determine the defective area in the measurement signal data and the range of the data portion that is replaced when it is determined to be defective are the same. However, different ranges may be used. As a specific example, the defective range removal unit 137 may replace the data portion of the measurement signal 1021 in a range wider than the range of the window 323 in which sudden noise is considered to have occurred with the data portion of the composite signal 1111 in that wider range.

[0116] Furthermore, the processing performed by the defective range removal unit 137 is also effective when performed on the data of one measurement signal for the channel P1 of interest. For this reason, in this embodiment, time series data having periodic (or nearly periodic) peaks for the signal of interest is used, but as another example, it may also be applied to time series data that does not have periodicity for the signal of interest.

[0117] <Exclusion process for channel data containing defects> For the sake of explanation, we will refer to Figure 9 to describe the process of excluding data from the measurement signal 1021 of the channel P1 of interest, which includes the defective area. In this example, we will explain the case where the area of ​​window 323 shown in Figure 9 is determined to be a defective area.

[0118] This example demonstrates a case where the measurement signal of the target channel P1 is acquired multiple times. Furthermore, this example shows a case where predetermined processing is performed based on these multiple measurement signals. Here, the prescribed process is not particularly limited and may be, for example, an averaging process. The averaging process may be, for example, an arithmetic averaging process.

[0119] The defective data exclusion unit 138 excludes the measurement signal data of the channel P1 of interest that includes the range determined to be defective when performing a predetermined process based on the measurement signal data of the channel P1 of interest multiple times. As a result, the measurement signal data of the channel P1 of interest that includes the range determined to be defective is excluded from the predetermined process and is not used in the predetermined process. In fields such as brain-related measurements, a single event in a recurring pattern is sometimes referred to as an epoch.

[0120] As a specific example, assuming that no defective range occurs, the process of averaging the data of the measurement signal of the channel P1 of interest over 100 measurements is performed. However, if a defective range occurs in the data of one of those 100 measurement signals, the processing unit 115 (for example, the defective data exclusion unit 138) may perform a process of averaging the data of the measurement signal of the channel P1 of interest over a total of 99 measurements. As another specific example, assuming that no defects occur, if 100 measurement signals of the channel P1 of interest are acquired and the data of these 100 measurement signals are averaged, then if a defect occurs in the data of one of those 100 measurement signals, the processing unit 115 (for example, the defective data exclusion unit 138) may acquire a total of 101 measurement signals of the channel P1 of interest and average the data of these 100 measurement signals.

[0121] Furthermore, the processing performed by the defective data exclusion unit 138 is applied when performing predetermined processing based on the data of the measurement signal of the channel of interest P1 multiple times. Therefore, it is suitable, for example, when time-series data having periodic (or nearly periodic) peaks for the signal of interest is used.

[0122] <Spike noise and block noise> In the information processing device 11 according to this embodiment, for example, sudden noise can be detected even when the sudden noise is spike noise or when the sudden noise is block noise, and the effects of the sudden noise can be suppressed.

[0123] Figure 12 shows an example of spike noise included in the measurement signal according to the embodiment. In the graph shown in Figure 12, the horizontal axis represents time, and the vertical axis represents level. Note that this graph is an illustrative example and does not necessarily represent a precise waveform. Figure 12 shows the measured signal 1311, as well as a schematic representation of the spike noise section 3311-3313, which is the part where spike noise occurs. Note that measurement signal 1311 is an example for illustrative purposes and may be considered as a group of measurement signals containing measurement signals from multiple channels, for example.

[0124] Figure 13 shows an example of block noise included in the measurement signal according to the embodiment. In the graph shown in Figure 13, the horizontal axis represents time, and the vertical axis represents level. Note that this graph is an illustrative example and does not necessarily represent a precise waveform. Figure 13 shows the measured signal 1411, as well as a schematic representation of the block noise sections 3411-3412, which are the parts where block noise occurs. Note that measurement signal 1411 is an example for illustrative purposes and may be considered as a group of measurement signals containing measurement signals from multiple channels, for example.

[0125] <Example of averaging> Refer to Figures 14 and 15 to see examples of the effects obtained by signal averaging. In the graphs shown in Figures 14 and 15, the horizontal axis represents time, and the vertical axis represents level. Note that these graphs are illustrative examples and do not necessarily represent precise waveforms.

[0126] Figure 14 shows an example of the effect of block noise when the measurement signal according to the embodiment is not averaged. Figure 14 shows the measurement signal 1511 of one channel of interest and the block noise 3511-3515 occurring in the measurement signal 1511. The measurement signal 1511 is a single measurement signal and is not an average of multiple measurement signals. In this example, block noise (3511-3515) is shown as an example of sudden noise, but other examples such as spike noise may also be used.

[0127] Figure 15 shows an example of suppressing the effect of block noise when the measurement signal according to the embodiment is averaged. Note that in the example in Figure 15, the scale of the horizontal axis (time axis) is different compared to the example in Figure 14. Figure 15 shows the average signal 1611 obtained by the processing according to this embodiment, the average signal without sudden noise 1612, and the average signal with sudden noise 1613.

[0128] The averaged signal 1613 with transient noise is the result of averaging (adding average) multiple measurement signals for the same channel of interest. In this case, at least one measurement signal contains transient noise (block noise in this example). In this example, artificial block noise is used as the transient noise.

[0129] The average signal 1612, free from transient noise, is the result of averaging (adding average) multiple measurement signals for the same channel of interest. In this case, all measurement signals used are those that do not contain transient noise (block noise in this example).

[0130] The averaged signal 1611 is the signal obtained by averaging the measurement signals of multiple channels of interest using the defective data exclusion unit 138 according to this embodiment, while excluding measurement signals that include the range determined to be defective (in this example, measurement signals that include block noise). In the example in Figure 15, the average signal 1611 approximates the waveform of the average signal 1612 without transient noise, and the effect of transient noise is more suppressed compared to the average signal 1613 with transient noise. In the example in Figure 15, the portion where transient noise is suppressed is shown as the noise section 3611-3615.

[0131] [Grid-type multi-channel sensor] Figure 16 shows an example of a sensor unit 611 having a grid-shaped multi-channel sensor according to a modified embodiment. Figure 16 shows the IJ coordinate system, a two-dimensional Cartesian coordinate system, for the sake of explanation. Alternatively, it may be called a matrix or similar structure instead of a grid.

[0132] The sensor unit 611 is equipped with (i × j) sensors B-11 to B-ij, where i is an integer greater than or equal to 2, and j is an integer greater than or equal to 2. Each of the (i × j) sensors B-11 to B-ij corresponds to a different channel.

[0133] (i × j) sensors B-11 to B-ij are arranged in a first predetermined interval of i units in a direction parallel to the I-axis, and in a second predetermined interval of j units in a direction parallel to the J-axis. The first predetermined interval is not particularly limited and may be, for example, equal intervals. The second predetermined interval is not particularly limited and may be, for example, equal intervals. Furthermore, the first predetermined interval and the second predetermined interval may be the same, for example.

[0134] In the example shown in Figure 1, a linear sensor unit 31 is used, but as another example of a configuration, a grid-shaped sensor unit 611 may be used, as shown in Figure 16. In such data with (i × j) channels, for example, there may be similarities between data from two adjacent channels, or between data from two or more adjacent channels.

[0135] Furthermore, in the grid-shaped sensor section 611, if we consider the case where i is 1 and j is multiple, or where i is multiple and j is 1, the sensor section becomes line-shaped. Furthermore, as another example of a configuration, a sensor unit having multiple sensors arranged in various configurations other than a line or grid may be used.

[0136] <Spatial arrangement of channel data> In the example shown in Figure 1, time-series data measured by a single sensor is used as the data for each channel. In this case, a time range is used as the predetermined window.

[0137] Here, as another configuration example, we show a case where the data for each channel is arranged according to its spatial configuration. Refer to Figure 16 for a specific example. Consider j sensors B-11 to B-1j arranged in a direction parallel to the J-axis. When the measurement signals (in this case, the values ​​at that time) from j sensors B-11 to B-1j at a given time are arranged according to the order in which these sensors B-11 to B-1j are located, data arranged according to their spatial configuration is generated.

[0138] Here, even when using a spatial axis (in this example, an axis parallel to the J-axis) instead of the time axis in the example of Figure 1, it can still serve as alternative data to the time-series data in the example of Figure 1. Therefore, instead of the time-series data in the example in Figure 1, data arranged according to spatial configuration may be used. In this case, a spatial range is used instead of a time range as the predetermined window range.

[0139] For example, the data corresponding to the spatial arrangement of j sensors B-11 to B-1j in the first direction parallel to the I-axis is used as data for one channel, and the data corresponding to the spatial arrangement of j sensors B-21 to B-2j in the second direction parallel to the I-axis is used as data for another channel, and so on, until data for i channels is obtained. In such i-channel data, for example, there may be similarities between two adjacent channels, or between two or more adjacent channels.

[0140] Here, considering j sensors arranged parallel to the J-axis (for example, sensors B-11 to B-1j), if the measurement signals of these j sensors at a given time (in this case, the values ​​at that time) are arranged according to the order in which these j sensors are positioned, the values ​​may be periodic in the direction of this arrangement. In other words, data arranged according to the spatial configuration may be periodic data. Furthermore, in data arranged according to spatial configuration, such periodicity may or may not be present.

[0141] Here, we have shown an example of data arranged according to spatial arrangement in a direction parallel to the J-axis. However, as another example, data arranged according to spatial arrangement in a direction parallel to the I-axis may be used, or data arranged according to spatial arrangement in other directions (for example, directions diagonal to the I-axis and J-axis) may be used.

[0142] For example, each of the sensors B-11 to B-ij may be an image sensor using a CCD or CMOS. Each of the sensors B-11 to B-ij may be used as a sensor that measures the signal of one pixel.

[0143] For example, if there are vertical and horizontal directions (for example, in the example in Figure 16, a direction parallel to the I-axis and a direction parallel to the J-axis), a group of multiple sensors arranged in a straight line in the vertical or horizontal direction may be considered as a sensor group for one channel. In this case, for example, processing as in this embodiment (processing such as determining the range of sudden noise) may be performed on two channels adjacent to the channel of interest (for example, channels adjacent in the vertical direction, or channels adjacent in the horizontal direction), or processing as in this embodiment may be performed on any number of channels (including all) adjacent to the channel of interest (for example, channels shifted in the vertical direction, or channels shifted in the horizontal direction).

[0144] [Regarding multiple sensors] Multiple sensors for measuring signals from multiple channels may be arranged in any configuration. For example, the sensors (or groups of sensors) for each channel may be placed in close proximity in space (with a small separation distance), or they may be placed in far distance in space (with a large separation distance). For example, multiple sensors may measure signals generated from the same signal source. In this embodiment, the measured signals of multiple channels are similar to (correlated with) each other. It has the characteristic that, for example, most of the waveforms of the measurement signals from multiple channels are similar. Such waveform similarity may be, for example, a similarity of waveforms that are different in terms of level. As a concrete example, there may be cases where, for optical signals generated from the same signal source, a sensor positioned on a predetermined side of the signal source (e.g., the east side) and a sensor positioned on a different side (e.g., the west side, south side, or north side) measure signals that are similar to each other.

[0145] In measurement signals from multiple channels, sudden noise can occur due to the individual conditions of each channel. Furthermore, sudden noise generated in the measurement signal of one channel does not affect the measurement signals of other channels. Therefore, the measurement signal of one channel where sudden noise occurs exhibits divergent behavior compared to the measurement signals of other channels due to the influence of that sudden noise. In this embodiment, for the sake of simplicity, we have shown a case where sudden noise occurs in the measurement signal of one of the multiple channels. However, it is also possible that sudden noise may occur simultaneously (or nearly simultaneously) in the measurement signals of two or more channels by chance.

[0146] In this embodiment, multiple sensors are positioned at different locations and are expected to acquire different information (measurement results). In other words, in this embodiment, the multiple sensors are not considered redundant sensors. The measurement signals from each channel are measurement signals from different locations in space, and are neither completely redundant nor completely independent. However, as another possible configuration, redundant sensors that are expected to acquire exactly the same information (measurement signals) from two or more sensors may be used as multiple sensors.

[0147] [Channels used for processing such as determining the range of sudden noise] When one of multiple channels is designated as the channel of interest, for example, the processing described in this embodiment (such as determining the range of sudden noise) may be performed using all other channels, or the processing described in this embodiment may be performed using only some of the other channels. Thus, when processing as in this embodiment is performed using some other channels, for example, the other channels with a higher similarity (lower dissimilarity) to the channel of interest may be preferentially used.

[0148] Furthermore, in this embodiment, after processing such as determining the range of sudden noise is performed on one channel as the channel of interest, similar processing is performed on other channels as the channel of interest, and so on, with each of the multiple channels being sequentially treated as the channel of interest and similar processing is performed. In this case, for example, for channels in which a defective range has already been determined to exist, the measurement signal data from which the defective range has been removed by the defective range removal unit 137 may be used, or the processing unit 115 (for example, the function of the defective data exclusion unit 138 may be used) may remove channels in which a defective range has been determined to exist from subsequent processing and not be used.

[0149] Furthermore, the method of sequentially treating each of the multiple channels as the channel of interest and performing the same processing is not necessarily used. As an alternative example, a method may be used in which two or more of the multiple channels are sequentially treated as the channel of interest and performed the same processing, or a method may be used in which one of the multiple channels is treated as the channel of interest and processed.

[0150] [Regarding measurement signals] The signal measured by the sensor (measured signal) can be a variety of signals. For example, the processing according to this embodiment may be applied to signals related to living organisms (biometric signals). The biological signal may be, for example, a bioelectrical signal or a biomagnetic signal. As a concrete example, the biosignal may be a measurement signal of the magnetic field originating from the human heart. For instance, it is possible to obtain information such as contour lines of the magnetic field using measurement signals from multiple MR sensors placed on the front (abdominal side) or side (arm side) of the human heart. Furthermore, for example, in a signal measuring heart rate, a periodic (or nearly periodic) peak due to the heart rate appears as a signal of interest.

[0151] Furthermore, although the example in Figure 1 shows a configuration in which the object 51 is moved by a belt conveyor 21, the movement of the object 51 may be carried out by a device other than the belt conveyor 21. Furthermore, while the example in Figure 1 shows a case where the relative position between the object 51 and the sensor unit 31 changes as the object 51 moves, other configuration examples include the sensor unit 31 being moved (for example, by rotation) instead of the object 51, or both the object 51 and the sensor unit 31 being moved.

[0152] Furthermore, the object being measured does not necessarily have to move during the measurement; for example, an object fixed in a predetermined position (a fixed position) may be used. The object to be measured for signal measurement may be, for example, an object that generates a signal periodically or approximately periodically. The signal may be, for example, an electric current signal, a magnetic field signal, an audio signal, or an vibration signal. As for the approximately periodic timing, for example, timing at time intervals that are close to periodic may be used. As a specific example, in a system that monitors the operation of a pump, which is the object of signal measurement, signals such as vibrations that occur periodically or approximately periodically due to the operation of the pump may be used as the signal to be measured.

[0153] Here, using Figure 16, we show an example where the object to be measured for signal measurement and the multi-channel sensor are each placed in a fixed position (a constant position). In this example, the sensor unit 611, which has a grid-like multi-channel sensor as shown in Figure 16, is positioned in a fixed location (a constant position). In this example, the object to be measured for signal measurement (not shown in Figure 16) is positioned at an arbitrary fixed position (a specific location) relative to the sensor unit 611. In such cases, if a periodic or nearly periodic signal originating from the object is measured by the sensor unit 611, a periodic or nearly periodic signal waveform can be measured, for example, as shown in Figures 4 to 7. Here, we show a case where a sensor unit 611 having a grid-shaped multi-channel sensor is used as the sensor unit to be placed in a fixed position. As the sensor unit 611, for example, a line-shaped sensor unit may be used when i is 1 and j is multiple, or when i is multiple and j is 1. In addition, as the sensor unit to be placed in a fixed position, a sensor unit having multiple sensors in various arrangements other than line-shaped or grid-shaped may be used.

[0154] Furthermore, while the example in Figure 1 shows a situation where measurement (signal measurement) by the sensor unit 31 is performed when the object 51 is being moved by the belt conveyor 21, for example, measurement (signal measurement) by the sensor unit 31 may also be performed when the belt conveyor 21 is stopped. In this case, if the object 51 that generates a periodic or approximately periodic signal is used, a periodic or approximately periodic signal waveform can be measured by the sensor unit 31, for example, as shown in Figures 4 to 7. Furthermore, in the example of Figure 1, a configuration may be used in which the object 51 is placed at an arbitrary fixed position (a certain position) without a belt conveyor 21. In this case, if the object 51 that generates a periodic or approximately periodic signal is used, a periodic or approximately periodic signal waveform can be measured by the sensor unit 31, for example, as shown in Figures 4 to 7.

[0155] As described above, in the information processing system 1 according to this embodiment, the information processing device 11 can accurately detect sudden noise contained in the target signal (in this embodiment, the measurement signal).

[0156] In the information processing device 11 according to this embodiment, the data of the channel of interest to be inspected is detected for sudden noise contained in the data of the channel of interest based on the degree of dissimilarity with the data of two or more other channels. Therefore, the information processing device 11 can detect the presence of sudden noise by suppressing the influence of the signal of interest, even when the data of the channel of interest is superimposed with the signal of interest.

[0157] In the information processing device 11 according to this embodiment, for example, even when a signal of interest that is included across the data of multiple channels and sudden noise specific to the channel of interest that should be removed are superimposed, it is possible to detect the sudden noise included in the data of the channel of interest. In the information processing device 11 according to this embodiment, for example, even when the magnitude of the sudden noise is about the same as the magnitude of the signal of interest, it is possible to suppress the influence of the signal of interest and detect the presence of the sudden noise. In the information processing device 11 according to this embodiment, it is possible to detect the time (or time period) when sudden noise occurs that occurs in a channel of interest.

[0158] Furthermore, when determining the presence or absence of transient noise in the data for each channel, it is difficult to accurately distinguish (separate) the signal of interest from the transient noise. For example, because transient noise occurs only rarely, methods that evaluate each channel using statistical values ​​such as variance tend to dilute the influence of transient noise, making it difficult to determine its presence or absence.

[0159] In the information processing device 11 according to this embodiment, for example, it is not necessary to prepare a simulated waveform of sudden noise in advance, and processing such as prior measurement is unnecessary, thereby simplifying the detection of sudden noise. Furthermore, in this embodiment, a wide range of signals can be used as the target signal for detecting sudden noise. In the information processing device 11 according to this embodiment, a good comparison target can be generated by adaptively synthesizing data from two or more other channels to determine the dissimilarity with the data of the channel of interest.

[0160] In the information processing device 11 according to this embodiment, the behavior of the data of the channel of interest can be explained to the greatest extent possible by the result of weighted synthesis of data from two or more channels other than the channel of interest (synthetic data), and any differences that could not be explained can be identified as dissimilar parts between the data of the channel of interest and the data of the other channels. Therefore, the information processing device 11 can determine the portion of the data of the channel of interest as a dissimilar portion, while not determining the portion of the data of the channel of interest that is common to the data of other channels as a dissimilar portion, and instead determining the portion of the sudden noise that is unique to the channel of interest as a dissimilar portion.

[0161] Here, the information processing device 11 can improve the accuracy of detecting sudden noise by weighting and combining the data of two or more channels other than the channel of interest, compared to, for example, simply adding or subtracting the data of two or more channels other than the channel of interest. Furthermore, when approximating the data of the channel of interest with the weighted composite data, computationally easy methods such as least-squares error approximation may be used.

[0162] In the information processing device 11 according to this embodiment, for example, a calculation method suitable for the type (characteristics) of sudden noise occurring in the data of multiple channels may be used as the dissimilarity calculation method. For example, when the dissimilarity is calculated using a value that corresponds to the sum of the absolute differences between the data of the channel of interest and the composite data, it is particularly easy to effectively reflect the magnitude of the sudden noise in the dissimilarity, especially when the sudden noise is blocky. For example, when the dissimilarity is calculated using a value that corresponds to the sum of the squared differences between the data of the channel of interest and the composite data, it is particularly easy to effectively reflect the magnitude of the sudden noise in the dissimilarity when the sudden noise is Gaussian in shape. For example, when the dissimilarity is calculated using a value corresponding to the maximum absolute difference between the data of the channel of interest and the composite data, it is particularly effective in reflecting the magnitude of the sudden noise when the sudden noise is spike-like. Such a dissimilarity can directly reflect, for example, the height of the sudden noise (the height of the spike).

[0163] In the information processing device 11 according to this embodiment, if there is a defective range in the data of the channel of interest, it is possible to replace only that defective range with a data portion based on data from another channel. Therefore, even if there is a defective range in the data of the channel of interest, the information processing device 11 can utilize the data portion other than the defective range, rather than discarding the entire data of the channel of interest. Here, when a defective range occurs due to sudden noise, the data portion other than the defective range is the normal data portion and often represents a large portion of the total data. In this way, the information processing device 11 can reduce the amount of missing data that occurs for a channel or time period (or spatial range), and remove the data portion of the defective range, making the clean data of the channel of interest available for subsequent processing.

[0164] In the information processing device 11 according to this embodiment, if there is a defective range in the data of the channel of interest, it is possible to discard the entire data of that channel of interest. Therefore, for example, when data is measured multiple times for the same channel of interest, it is possible to remove data with defects and perform various processing using clean data.

[0165] In the information processing device 11 according to this embodiment, it is possible to set a threshold used for determining the dissimilarity of the data of the channel of interest based on the data of multiple channels. Therefore, the information processing device 11 can detect sudden noise specific to the data of the channel of interest, without being affected by, for example, the steady-state noise environment in the sensor of the channel of interest or the characteristics of the target signal.

[0166] In the information processing device 11 according to this embodiment, the width of the window for detecting sudden noise occurring in the data of the channel of interest can be set to the width of one sample. Therefore, the information processing device 11 can effectively detect sudden noises with a narrow duration, and can accurately identify the occurrence time (occurrence time) of sudden noises.

[0167] <Configuration example relating to the above embodiment> As an example configuration, the program (in this embodiment, the program executed by the processor in the information processing device 11) is a program that causes the computer to implement the following: an acquisition function (in the example of Figure 2, the function of the acquisition unit 131) that acquires data from three or more channels; a selection function (in the example of Figure 2, the function of the selection unit 132) that selects data from a single channel as the channel of interest from the data of multiple channels acquired by the acquisition function; and a dissimilarity calculation function (in the example of Figure 2, the function of the dissimilarity calculation unit 134) that calculates the dissimilarity of the data from two or more channels that were not selected by the selection function for each predetermined range of the channel of interest data selected by the selection function.

[0168] As an example configuration, the program further includes a function to enable the computer to generate composite data, which is the result of combining data from two or more channels that have not been selected by the selection function (in the example in Figure 2, this is the function of the composite data generation unit 133). The dissimilarity calculation function calculates the dissimilarity of the composite data generated by the composite data generation function.

[0169] As an example configuration, in the program, the synthetic data generation function generates synthetic data by adjusting the weights for combining data from two or more channels that were not selected by the selection function, using least-squares error approximation on the channel data of interest.

[0170] As one example configuration, in the program, the dissimilarity calculation function calculates dissimilarity based on the sum of the absolute differences between the channel data of interest and the composite data. As one example configuration, in the program, the dissimilarity calculation function calculates dissimilarity based on the sum of the squared differences between the channel data of interest and the composite data. As one example configuration, in the program, the dissimilarity calculation function calculates dissimilarity based on the maximum absolute value of the difference between the channel data of interest and the composite data.

[0171] As one example configuration, in the program, the dissimilarity calculation function calculates dissimilarity based on the difference between the features of the target channel data and the features of the composite data. As one example configuration, the program's dissimilarity calculation function calculates dissimilarity based on the difference between the probability distribution of the channel data of interest and the probability distribution of the composite data. As one example configuration, in the program, the dissimilarity calculation function calculates dissimilarity based on the difference between the power spectrum of the channel data of interest and the power spectrum of the composite data. As one example configuration, in the program, the dissimilarity calculation function calculates dissimilarity using a calculation formula that associates samples from the channel data of interest with the composite data.

[0172] As an example configuration, the program further includes a dissimilarity determination function (the function of the dissimilarity determination unit 135 in the example of Figure 2) that determines whether the dissimilarity calculated by the dissimilarity calculation function is higher than a predetermined threshold, and a defect range determination function (the function of the defect range determination unit 136 in the example of Figure 2) that determines a predetermined range in which the dissimilarity determination function determines to be higher than the threshold as a defect range.

[0173] As an example configuration, the program further includes a program that enables the computer to implement a defective range removal function (the function of the defective range removal unit 137 in the example of Figure 2), which replaces the data portion of a predetermined range determined as a defective range by the defective range determination function with the data portion resulting from the synthesis of data from two or more channels that were not selected by the selection function.

[0174] As an example configuration, the program further implements a defective data exclusion function (the function of the defective data exclusion unit 138 in the example of Figure 2) that, when performing predetermined processing based on multiple instances of the target channel data, excludes the target channel data that includes a predetermined range determined as a defective range by the defective range determination function.

[0175] As an example configuration, the program further implements a threshold determination function (the function of the threshold determination unit 139 in the example of Figure 2) that enables the computer to determine a threshold based on the statistical results of the dissimilarity of each range of the channel data of interest, using the dissimilarity obtained for each range that is the same as or different from a predetermined range.

[0176] As one example of a configuration, in a program, the specified range is two or more different ranges. As one example configuration, in the program, the predetermined range is the range of one sample of the channel data of interest. As an example configuration, in the program, the data from multiple channels is time-series data. Furthermore, the specified range is a time period. As an example configuration, in the program, the data for multiple channels is data arranged according to its spatial configuration. And the predetermined range is a spatial range.

[0177] For example, we can also provide information processing equipment. As an example configuration, the information processing device (information processing device 11 in the example of Figure 1) comprises: an acquisition unit (acquisition unit 131 in the example of Figure 2) that acquires data from three or more channels; a selection unit (selection unit 132 in the example of Figure 2) that selects data from a single channel as the channel of interest from the data of multiple channels acquired by the acquisition unit; and a dissimilarity calculation unit (dissimilarity calculation unit 134 in the example of Figure 2) that calculates the dissimilarity of the data from two or more channels that were not selected by the selection unit for each predetermined range of the channel of interest selected by the selection unit.

[0178] For example, we can also provide information processing methods. As an example configuration, in the information processing method (in the example in Figure 1, the method performed in the information processing device 11), the acquisition unit of the information processing device acquires data from three or more channels, the selection unit of the information processing device selects data from a single channel from the data of multiple channels acquired by the acquisition unit as the channel data of interest, and the dissimilarity calculation unit of the information processing device calculates the dissimilarity of the channel data of interest selected by the selection unit for each predetermined range of the channel data of interest, with respect to data from two or more channels that were not selected by the selection unit.

[0179] For example, we can also provide information processing systems. As an example configuration, the information processing system (information processing system 1 in the example of Figure 1) comprises a sensor unit (sensor unit 31 in the example of Figure 1) that measures data from three or more channels, and an information processing device (information processing device 11 in the example of Figure 1). The information processing device includes an acquisition unit that acquires data from multiple channels measured by a sensor unit, a selection unit that selects data from a single channel as the focus channel data from the data from multiple channels acquired by the acquisition unit, and a dissimilarity calculation unit that calculates the dissimilarity of the focus channel data selected by the selection unit for each predetermined range of the focus channel data with respect to data from two or more channels that were not selected by the selection unit.

[0180] Furthermore, a program to realize the function of any component in any of the devices described above may be recorded on a computer-readable recording medium, and that program may be loaded into a computer system and executed. Here, "computer system" includes hardware such as operating systems and peripheral devices. "Computer-readable recording medium" refers to portable media such as flexible disks, magneto-optical disks, ROMs, CD (Compact Disc)-ROMs (Read Only Memory), and storage devices such as hard disks built into computer systems. Moreover, "computer-readable recording medium" also includes volatile memory within a computer system that retains a program for a certain period of time, such as volatile memory within a computer system that acts as a server or client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line. Such volatile memory may be, for example, RAM (Random Access Memory). The recording medium may be, for example, a non-temporary recording medium.

[0181] Furthermore, the above program may be transmitted from a computer system that stores the program in a memory device or the like to another computer system via a transmission medium or by transmission waves within the transmission medium. Here, the "transmission medium" used to transmit the program refers to a medium that has the function of transmitting information, such as a network like the Internet or a communication line like a telephone line. Furthermore, the above program may be intended to implement only a portion of the functions described above. Moreover, the above program may be a so-called differential file, capable of implementing the aforementioned functions in combination with programs already recorded in the computer system. A differential file may also be called a differential program.

[0182] Furthermore, the functions of any component in any device described above may be implemented by a processor. For example, each process in the embodiment may be implemented by a processor that operates based on information such as a program, and a computer-readable recording medium that stores information such as a program. Here, the processor may be implemented by having the functions of each part implemented by separate hardware, or by having the functions of each part implemented by integrated hardware. For example, the processor includes hardware, and the hardware may include at least one of a circuit that processes digital signals and a circuit that processes analog signals. For example, the processor may be configured using one or more circuit devices or one or both of one or more circuit elements mounted on a circuit board. An IC (Integrated Circuit) may be used as the circuit device, and a resistor or capacitor may be used as the circuit element.

[0183] Here, the processor may be, for example, a CPU. However, the processor is not limited to a CPU; various types of processors may be used, such as a GPU (Graphics Processing Unit) or a DSP (Digital Signal Processor). The processor may also be, for example, a hardware circuit using an ASIC (Application Specific Integrated Circuit). Furthermore, the processor may be composed of, for example, multiple CPUs, or multiple hardware circuits using ASICs. The processor may also be composed of, for example, a combination of multiple CPUs and multiple hardware circuits using ASICs. The processor may also include, for example, one or more amplifier circuits or filter circuits that process analog signals.

[0184] While embodiments of this disclosure have been described in detail above with reference to the drawings, the specific configuration is not limited to these embodiments and may include designs and other elements that do not depart from the gist of this disclosure. [Explanation of Symbols]

[0185] 1...Information processing system, 11...Information processing device, 12...A / D converter, 13...Display device, 21...Belt conveyor, 31, 611...Sensor unit, 41...Unit, 42...Current supply unit, 51...Object, 61...Rotation center axis, 62...Rotation direction, 111...Input unit, 112...Output unit, 113...Communication unit, 114...Storage unit, 115...Processing unit, 131...Acquisition unit, 132...Selection unit, 133...Synthetic data generation unit, 134...Dissimilarity calculation unit, 135...Dissimilarity determination unit, 136...Defective range determination unit, 137...Defective range removal unit, 138...Defective data exclusion unit, 139...Threshold determination unit, 311, 321...Lower limit, 312, 322...Upper limit, 313, 323...Window, 351, 352...Comparison, 361~363...Movement direction, 411, 521...Threshold, 431, 4 32...Sudden noise area, 511...Normal noise area, 512...Sudden noise area, 1011, 1021, 1031, 1041, 1051, 1311, 1411, 1511...Measured signal, 1111...Composite signal, 1211...Dissimilarity characteristics, 1611...Average signal, 1612...Average signal without sudden noise, 1613...Average signal with sudden noise, 2011, 2021, 2 022...Sudden noise, 3011~3015, 3031~3035, 3041~3045, 3051~3055, 3111~3115...Peak position, 3311~3313...Spike noise section, 3411, 3412...Block noise section, 3511~3515...Block noise, 3611~3615...Noise section, A-1~An, B-11~B-ij...Sensor

Claims

1. A data acquisition function that acquires data from three or more channels, A selection function that selects data from a single channel as the focus channel data from the data of multiple channels acquired by the acquisition function, A dissimilarity calculation function calculates the dissimilarity of the data of two or more channels among the channels not selected by the selection function for each predetermined range of the channel data of interest selected by the selection function, A composite data generation function that generates composite data which is the result of combining data from two or more channels that are not selected by the selection function, A program to make a computer realize this, The dissimilarity calculation function calculates the dissimilarity to the composite data generated by the composite data generation function, The composite data generation function generates the composite data by adjusting the weights for combining data from two or more channels that have not been selected by the selection function, so that the composite data is similar to the data of the channel of interest. program.

2. The composite data generation function generates the composite data by adjusting the weights for compositing data from two or more channels that were not selected by the selection function, using the least squares error approximation of the channel data of interest. The program according to claim 1.

3. The dissimilarity calculation function calculates the dissimilarity based on the sum of the absolute differences between the channel data of interest and the composite data. The program according to claim 1 or claim 2.

4. The dissimilarity calculation function calculates the dissimilarity based on the sum of the squares of the differences between the channel data of interest and the composite data. The program according to claim 1 or claim 2.

5. The dissimilarity calculation function calculates the dissimilarity based on the maximum absolute value of the difference between the channel data of interest and the composite data. The program according to claim 1 or claim 2.

6. The dissimilarity calculation function calculates the dissimilarity based on the difference between the feature quantities of the channel data of interest and the feature quantities of the composite data. The program according to claim 1 or claim 2.

7. The dissimilarity calculation function calculates the dissimilarity based on the difference between the probability distribution of the channel data of interest and the probability distribution of the composite data. The program according to claim 1 or claim 2.

8. The dissimilarity calculation function calculates the dissimilarity based on the difference between the power spectrum of the channel data of interest and the power spectrum of the composite data. The program according to claim 1 or claim 2.

9. The dissimilarity calculation function calculates the dissimilarity using a calculation formula that associates samples from the channel data of interest with the composite data. The program according to claim 1 or claim 2.

10. moreover, A dissimilarity determination function that determines whether the dissimilarity calculated by the dissimilarity calculation function is higher than a predetermined threshold, A defect range determination function that determines the predetermined range in which the dissimilarity determination function determines that the dissimilarity is higher than the threshold is a defect range, The program is for the computer to implement the above. The program according to any one of claims 1 to 9.

11. moreover, The program is for enabling the computer to perform a defective range removal function, which replaces the data portion of the predetermined range determined as a defective range by the defective range determination function with the data portion resulting from the synthesis of data from two or more channels among the channels not selected by the selection function. The program according to claim 10.

12. moreover, The program is for enabling the computer to implement a defective data exclusion function that excludes the defective channel data including the predetermined range determined as a defective range by the defective range determination function when performing predetermined processing based on the aforementioned defective channel data multiple times. The program according to claim 10.

13. moreover, The program is for implementing a threshold determination function in the computer, which determines the threshold based on the statistical results of the dissimilarity of each of the ranges of the channel data of interest, using the dissimilarity obtained for each range that is the same as or different from the predetermined range. The program according to any one of claims 10 to 12.

14. The aforementioned predetermined range is two or more different ranges. The program according to any one of claims 1 to 13.

15. The predetermined range is the range of one sample of the channel data of interest. The program according to any one of claims 1 to 14.

16. The data from the aforementioned multiple channels are each time-series data. The aforementioned predetermined range is a time period range. The program according to any one of claims 1 to 15.

17. The data in the aforementioned multiple channels are, in each case, data arranged according to their spatial configuration. The aforementioned predetermined range is a spatial range. S according to any one of claims 1 to 16.

18. An acquisition unit that acquires data from three or more channels, A selection unit selects data from a single channel as the channel of interest data from the data of the multiple channels acquired by the acquisition unit, A dissimilarity calculation unit calculates the dissimilarity of the data of two or more channels among the channel data not selected by the selection unit for each predetermined range of the channel data of interest selected by the selection unit, A composite data generation unit generates composite data which is the result of combining data from two or more channels that are not selected by the selection unit. Equipped with, The dissimilarity calculation unit calculates the dissimilarity to the composite data generated by the composite data generation unit, The composite data generation unit generates the composite data by adjusting the weights for combining data from two or more channels that have not been selected by the selection unit, so that the composite data is similar to the data of the channel of interest. Information processing device.

19. The data acquisition unit of the information processing device acquires data from three or more channels. The selection unit of the information processing device selects the data of a single channel from the data of the multiple channels acquired by the acquisition unit as the channel data of interest. The dissimilarity calculation unit of the information processing device calculates the dissimilarity of the data of two or more channels among the channel data not selected by the selection unit for each predetermined range of the channel data of interest selected by the selection unit. The composite data generation unit of the information processing device generates composite data which is the result of combining data from two or more channels that were not selected by the selection unit. The dissimilarity calculation unit calculates the dissimilarity to the composite data generated by the composite data generation unit, The composite data generation unit generates the composite data by adjusting the weights for combining data from two or more channels that have not been selected by the selection unit, so that the composite data is similar to the data of the channel of interest. Information processing methods.

20. An information processing system comprising a sensor unit that measures data from three or more channels, and an information processing device, The aforementioned information processing device is An acquisition unit that acquires data from the plurality of channels measured by the sensor unit, A selection unit selects data from a single channel as the channel of interest data from the data of the multiple channels acquired by the acquisition unit, A dissimilarity calculation unit calculates the dissimilarity of the data of two or more channels among the channel data not selected by the selection unit for each predetermined range of the channel data of interest selected by the selection unit, A composite data generation unit generates composite data which is the result of combining data from two or more channels that are not selected by the selection unit. Equipped with, The dissimilarity calculation unit calculates the dissimilarity to the composite data generated by the composite data generation unit, The composite data generation unit generates the composite data by adjusting the weights for combining data from two or more channels that have not been selected by the selection unit, so that the composite data is similar to the data of the channel of interest. Information processing system.

Images