Automatic steering systems for ships

Abstract

This provides a technique for calculating updated values ​​that include the effects of disturbances. [Solution] One embodiment is a ship's automatic steering system that outputs a command rudder angle using hull parameters based on a reference bearing and a bow bearing, and comprises an identification calculation unit that calculates an identification value by adjusting the hull parameters such that an evaluation quantity, which is the sum of squared errors between the model output data and the output data, which is an actual measured value relating to the hull, is minimized during an identification value calculation period that includes at least the start time of the hull's change of course, and an update calculation unit that calculates coefficients of an update function, which is a regression equation relating to the hull's speed, based on an identification data sequence in which the calculated identification value and the evaluation quantity corresponding to the identification value are accumulated in time series, and calculates updated values ​​of the hull parameters using the update function to which the calculated coefficients are given.

Description

[Technical Field]

[0001] This invention relates to a technique for identifying hull parameters in an identification model. [Background technology]

[0002] A ship's automatic steering system is a device that controls the rudder to make the heading from a gyrocompass follow a set course. Its control system calculates the deviation and turning angular velocity from the input of the set course and the ship's heading, multiplies them by a control gain, and outputs the command rudder angle, which is the control variable, to the steering gear. The steering gear moves the rudder to induce a turning angular velocity in the hull and change the heading. More specifically, the command rudder angle is calculated by adding the output of the feedback controller, which multiplies the deviation by a control gain, and the output of the feedforward controller. Identified hull parameters are input to the trajectory calculation unit, which calculates the reference heading based on the trajectory plan from the set course, the feedback controller, and the feedforward controller, and these hull parameters are used in the calculations and controls of the trajectory calculation unit, feedback controller, and feedforward controller.

[0003] Hull parameters are the parameters that constitute the hull motion model controlled by a ship's automatic steering system. Since these hull parameters are unknown in most cases, they are obtained through parameter identification. For example, in ships such as cargo ships and tankers, the draft of the hull changes due to loading and unloading of cargo, which alters the hull characteristics. Therefore, if control gains based on the hull parameters in the unloaded state are used for a hull loaded with cargo, the closed-loop stability of the steering system may decrease, potentially leading to a yawing phenomenon. To avoid such a situation, ship's automatic steering systems identify the hull parameters. In other words, ship's automatic steering systems improve the controllability of the hull by appropriately identifying these hull parameters.

[0004] As a technique related to such parameter identification, it includes an identification calculation unit that calculates an identified value of the hull parameters, and an update calculation unit that calculates an updated value based on the identified value. The identification calculation unit includes an identification model including a hull model related to the hull motion of the hull, a parameter adjustment unit that calculates an identified value obtained by adjusting the hull parameters from the comparison result between the model output data and the output data which is the measured value related to the hull, and an average ship speed calculation unit that calculates the average value of the surge speed of the hull as the average ship speed corresponding to the identified value. The update calculation unit includes an acquisition unit that acquires the identified value and the average ship speed, an identified value storage unit that accumulates the acquired identified values in time series, a calculation processing unit that calculates the coefficient of an update function which is a linear function related to the ship speed by the least squares method based on the accumulated identified values, and calculates an updated value of the hull parameters by the update function using the calculated coefficient and the acquired average ship speed. A ship autopilot device characterized by this is known (see Patent Document 1).

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Non-Patent Documents

[0006]

Non-Patent Document 1

Non-Patent Document 2

Non-Patent Document 3

Summary of the Invention

Problems to be Solved by the Invention

[0007] In the above-mentioned conventional automatic steering device for ships, the hull parameter identification system estimates the identified values of hull parameters from the hull motion and estimates updated values from the estimated identified values. The estimated updated values become the nominal values of the control system. In such estimation of updated values, the regression equation of hull parameters is characterized by the least squares method with respect to the ship speed, and outliers are removed from the regression equation.

[0008] The identified values can be affected by disturbances, and the outliers to be removed may include the influence of disturbances. Furthermore, the least squares method is strongly affected by the presence or absence of outliers. Therefore, simply removing outliers from the regression equation will exclude the influence of disturbances from the estimated updated values, and the control system will use inappropriate updated values.

[0009] The present invention has been made to solve the above-mentioned problems, and an object thereof is to provide a technique for calculating updated values that include the influence of disturbances.[[ID=第十九]]

Means for Solving the Problems

[0010] One embodiment is a ship's automatic steering system that outputs a command rudder angle using hull parameters based on a reference bearing and a bow bearing, and includes an identification model that includes a hull model relating to the hull motion of the ship, which outputs model output data from predetermined input data obtained by the ship's automatic steering system, and calculates an identification value by adjusting the hull parameters such that an evaluation quantity, which is the sum of squares of errors between the model output data and output data that are measured values ​​relating to the hull, is minimized during an identification value calculation period that includes at least the start of the ship's change of course, and an update calculation unit that calculates coefficients of an update function, which is a regression equation relating to the ship's speed of the ship, based on an identification data sequence in which the calculated identification value and the evaluation quantity corresponding to the identification value are accumulated in time series, and calculates updated values ​​of the hull parameters using the update function to which the calculated coefficients are given. [Effects of the Invention]

[0011] According to the present invention, a technique is provided that can calculate an updated value that includes the effects of disturbances. [Brief explanation of the drawing]

[0012] [Figure 1] This is a block diagram showing a control system including an automatic steering device for a ship according to an embodiment. [Figure 2] This figure shows the statistical characteristics of the identification data. [Figure 3] This figure shows the Biwave function and the normal distribution. [Figure 4] This figure shows the update function and the identification data. [Figure 5] This flowchart shows the operation of the update calculation process. [Figure 6] This is a flowchart showing the operation of a subroutine. [Figure 7] This figure shows the results of outlier removal. [Figure 8] This is a diagram showing the evaluation quantity and its weight. [Figure 9] This table shows the results of the updated values. [Figure 10]This figure shows the turning force gain Kr as an updated value using the conventional method. [Figure 11] This figure shows the turning force gain Kr as an updated value according to the embodiment. [Figure 12] This figure shows the lateral flow gain Kv as the updated value using the conventional method. [Figure 13] This figure shows the lateral flow gain Kv as an updated value according to the embodiment. [Figure 14] This figure shows the time constant Tr as the updated value using the conventional method. [Figure 15] This figure shows the time constant Tr as the update value according to the embodiment. [Figure 16] This figure shows the time constant Tr3 as the updated value using the conventional method. [Figure 17] This figure shows the time constant Tr3 as the update value according to the embodiment. [Modes for carrying out the invention]

[0013] Embodiments of the present invention will be described below with reference to the drawings.

[0014] 1. Configuration of a ship's automatic steering system First, a control system including a ship's automatic steering system according to the present invention will be described. Figure 1 is a block diagram showing a control system including a ship's automatic steering system according to an embodiment.

[0015] As shown in Figure 1, the control system in this embodiment includes a ship's automatic steering system 1 and the ship 2 that it controls. The ship 2 comprises a steering gear 21, a hull 22, and sensors 23. The ship's automatic steering system 1 comprises a trajectory calculation unit 11, an identification calculation unit 12, an update calculation unit 13, a subtractor 14, a feedback controller 15, a feedforward controller 16, and an adder 17.

[0016] The orbit calculation unit 11 calculates the input set course ψ S Reference bearing ψ based on orbital plan RIt performs calculations. The identification calculation unit 12 identifies hull parameters, which are parameters constituting the hull model described later, and outputs them as the identified values of the hull parameters. The update calculation unit 13 calculates the updated values of the hull parameters based on a plurality of identified values identified by the identification calculation unit 12, and outputs these updated values to the trajectory calculation unit 11, the feedback controller 15, and the feedforward controller 16. The trajectory calculation unit 11, the feedback controller 15, and the feedforward controller 16 perform respective calculations and controls using the updated values of the hull parameters.

[0017] The subtractor 14 outputs the deviation e between the reference azimuth ψ output from the trajectory calculation unit 11 R and the bow azimuth ψ of the hull 22. The feedback controller 15 multiplies the deviation e output from the subtractor 14 by a control gain to obtain the feedback rudder angle δ FB and outputs it. The feedforward controller 16 outputs the feedforward rudder angle δ R based on the reference azimuth ψ output by the trajectory calculation unit 11. FF The adder 17 adds the feedback rudder angle δ output by the feedback controller 15 FB and the feedforward rudder angle δ output by the feedforward controller 16 FF to output the commanded rudder angle δ C to the steering machine 21.

[0018] Also, the sensors 23 of the ship 2 include a speed log that detects the surge speed (hereinafter referred to as the ship speed) u of the hull 22, a gyrocompass that detects the bow azimuth ψ of the hull 22, and a GNSS sensor that detects the hull position (x, y) from a satellite positioning system (GNSS) such as GPS. The ship speed u, the bow azimuth ψ, and the hull position (x, y) detected by the sensors 23 are input to the identification calculation unit 12, and the ship speed u and the hull position (x, y) detected by the sensors 23 are input to the update calculation unit 13. Here, the hull position (x, y) is the hull position in the local horizontal coordinate system where the X-axis is northward and the Y-axis is eastward.

[0019] 2 Identification System The identification system according to this embodiment will be described.

[0020] The identification system provides adaptive functionality to the control system and consists of an identification calculation unit 12 and an update calculation unit 13. It outputs the parameters of the hull model and the wave model as nominal values. In the control system, these nominal values ​​are used to calculate setpoints and control gains.

[0021] The parameters of the hull model are determined by an identification process that identifies them from time-series data of measurements taken by sensors 23 during course changes. This identification process includes an identification calculation process by an identification calculation unit 12 and an update calculation process by an update calculation unit 13. This embodiment relates to the update calculation process; for details of the identification calculation process, please refer to Patent Document 1.

[0022] The estimation of updated values ​​based on the identified values ​​calculated by the identification calculation process is performed by a regression process and a parameter calculation process. Here, the regression process is the process of finding the regression equation from the scatter plot, and the parameter calculation process is the process of finding the hull parameters corresponding to the ship based on the regression process.

[0023] 2.1 Statistical Assumptions for Identification Values The assumptions regarding the statistical characteristics of the identification values ​​according to this embodiment will now be explained. Figure 2 shows the statistical characteristics of the identification data.

[0024] In this embodiment, the identification value is obtained by an identification calculation process that minimizes the evaluation quantity. The evaluation quantity consists of the sum of the squares of the errors between the output of the target model and the detected quantity, and ideally it is zero. However, in reality, the evaluation quantity has errors due to modeling errors, course change conditions, sea conditions, etc. At this time, the statistical characteristics of the identification data, which includes at least one type of identification value as shown in Figure 2, are assumed to be as follows. In Figure 2, N is a normal distribution, μ is the population mean, and σ is the population variance.

[0025] The identified values ​​are derived for each change in needle position and represent independent events following a normally distributed population (parameter). In the parameter, the population mean is unknown, but the population variance can be approximated by a known evaluation quantity. • Individual identification values ​​are obtained from the reciprocals or combinations of variables, and in doing so, they are associated with the evaluation quantity. The evaluation quantity includes identification errors such as wave components. A smaller evaluation quantity results in higher reliability of the identification value. Conversely, a larger evaluation quantity results in lower reliability of the identification value.

[0026] 2.2 Identification Dataset The identification data D of the hull parameters calculated by the identification calculation process

[0027]

number

[0028] Identification data D is data column D i The data is accumulated in a time series. Here, the subscript i is i=1,2,...,n, where n is the number of data points. Identification data sequence D i Updated value

[0029]

number

[0030] 3. Regression Processing This section explains the regression process for calculating the coefficients of the regression equation.

[0031] In this embodiment, the regression equation is

[0032]

number

[0033] 3.1 Introduction of Weights When the least squares method is derived based on the maximum likelihood method (see Non-Patent Document 1), variance is also assigned. By converting the variance into weights, the weighted least squares method can be obtained.

[0034] When y is observed according to a normal distribution of the population (parameter), the probability of observing y^ is

[0035]

number

[0036] According to the maximum likelihood method, the likelihood function (likelihood) L is expressed as two variables for simplicity of representation.

[0037]

number

[0038] If we take out the exponent part of the above equation,

[0039]

number

[0040] Therefore, if we substitute the reciprocal of the variance with the weight w (bold),

[0041]

number

[0042]

number

[0043] Therefore, by introducing weights, the certainty and error of the data can be taken into account in the estimate. In other words, if the weight is large (the variance is small), the influence of the data corresponding to that weight on the estimate increases. Conversely, if the weight is small, the influence of the data corresponding to that weight on the estimate decreases. Furthermore, if the weight is zero, the data corresponding to that weight has no influence on the estimate at all.

[0044] Once again, the weights related to the evaluation quantity in equation (7)

[0045]

number

[0046]

number

[0047] Therefore, by assigning weights to the identified values, the accuracy of estimating the updated values ​​can be improved. From now on, we will use weights in our descriptions.

[0048] 3.2 Outlier Removal In the identification system according to this embodiment, outliers are removed by weights. When the identification system detects an outlier in the data, it reduces the influence of the outlier by decreasing or setting the corresponding weight to zero. Therefore, the weight becomes the product of the weight of the evaluation quantity and the weight of the outlier.

[0049]

number

[0050]

number

[0051] Several methods have been proposed for outlier removal by applying a filtering function to weights (see Non-Patent Document 2). In this embodiment, the weights for outlier removal are determined using Tukey's Biweight method (see Non-Patent Document 3).

[0052]

number

[0053] First, we calculate the estimation error ε, excluding the outlier removal weights, using the following formula.

[0054]

number

[0055] Next, we derive the outlier deviation d from the following formula.

[0056]

number

[0057]

number

[0058] Figure 3 shows the Biweight function and the normal distribution. As shown in Figure 3, the standard normal distribution N(0,1 2 The probability of ) is 68.3% at ±1s, 86.7% at ±1.5s, and 95.5% at ±2s. Therefore, in this embodiment, the setting range is,

[0059]

number

[0060] 3.3 Regression Coefficients In this embodiment, the regression coefficients are obtained by partially differentiating equation (7) (extended to the number of data points n) with respect to β0 and β1, respectively, and setting them to zero. In this algorithm, the direct method and the gradient method are used. These will be the elements that constitute the update calculation process, as will be described later.

[0061] 1. The direct method directly calculates the regression coefficients. 2. The gradient method uses convergence calculations to remove outliers.

[0062] 3.3.1 Direct method The direct method calculates the coefficients of the regression equation from the data.

[0063] For a 0th-degree polynomial: If we partially differentiate equation (6) with respect to the coefficient β0 and set it to zero,

[0064]

number

[0065]

number

[0066] For a linear equation: Substituting the coefficients β0 and β1 into the above,

[0067]

number

[0068]

number

[0069]

number

[0070] 3.3.2 Gradient Method The direct method has difficulty handling outlier removal. Estimated error ε i =w i (y^ i -y i Since ) is not used, outlier deviation d i This is because it is not possible to set it. Therefore, the gradient method is used to handle outlier removal. In the gradient method, the regression coefficients are found by searching for the minimum value from the estimation error of the regression equation. In this process, the outlier deviation d i We find the weights for outlier removal w Oi This is set. Therefore, the gradient method requires convergence calculation.

[0071] The likelihood function can be derived from equation (6).

[0072]

number

[0073] Furthermore, the coefficient η that determines the degree of convergence is

[0074]

number

[0075] η is called the coefficient of determination; a large value indicates divergence, while a small value indicates difficulty in convergence. If it is difficult to set η, the hull parameters can be made dimensionless.

[0076] For a zero-degree polynomial: When equation (25) is partially differentiated with respect to the coefficient β0, the gradient coefficient is

[0077]

number

[0078]

number

[0079] In the case of a linear expression: Similarly, if we take partial derivatives with respect to coefficients β0 and β1, the gradient coefficient is

[0080]

number

[0081]

number

[0082] 4. Update calculation process The update calculation process for determining the updated values ​​of the hull parameters, as described above, will now be explained. Figure 4 shows the update function and the identification data.

[0083] In Figure 4, P3 is an outlier and has been removed. The update function represents the regression equation.

[0084] Update value P^

[0085]

number

[0086] 4.1 Response to ship speed We will consider cases where the ship speed range is narrow and cases where it is wide.

[0087] 4.1.1 When the ship speed range is narrow When the ship speed range is narrow, this approach addresses situations where the number of ship speed datasets is small or where the ship speed datasets are concentrated within a narrow ship speed range. In this case, a zero-degree regression equation is used instead of a first-degree one. This is because first-degree equations tend to have steep slopes. Using a zero-degree equation stabilizes the coefficients of the regression equation and suppresses fluctuations in the estimated values.

[0088] 4.1.2 When the ship speed range is wide This approach is suitable when the ship speed dataset covers a wide range of ship speeds. In this case, a linear regression equation is used. That is, the regression equation is expressed in terms of slope and intercept.

[0089] 4.1.3 Setting the ship speed range Ship speed range Δ of identification data u From the ship speed dataset u

[0090]

number

[0091]

number

[0092] The ship speed range is linked to the ship speed in the identification data. The degree of the regression equation

[0093]

number

[0094]

number

[0095] 4.2 Calculation of the update function This section explains the procedure for calculating an update function (regression equation) corresponding to ship speed using the update calculation process and its subroutines.

[0096] 4.2.1 Update Calculation Process The operation of the update calculation process according to this embodiment will now be described. Figure 5 is a flowchart showing the operation of the update calculation process. Note that the operation shown in Figure 5 is the operation that occurs immediately after the control system starts up, and the update calculation process is performed at predetermined intervals.

[0097] As shown in Figure 5, first, the update calculation unit 13 acquires the identification dataset and calculation conditions (S101) and calculates the initial values ​​of the regression coefficients (S102). Here, the update calculation unit 13 calculates the initial values ​​using the direct method described above in Section 3.3.1. Here, it is desirable to use a zero-degree equation to obtain approximate values ​​as initial values. Note that if the update calculation process being executed is not performed immediately after the start of operation of the control system, the calculation of initial values ​​will be omitted.

[0098] Next, the update calculation unit 13 calculates the ship speed range Δ u It is determined whether the value is above the range threshold (S103).

[0099] ​If the ship speed range Δu is greater than or equal to the range threshold (S103, YES), the update calculation unit 13 sets the regression equation to a linear equation (S104), executes a subroutine process described later (S105), stores the regression coefficients in a memory device (not shown) or the like so that they can be used in the next cycle's update calculation process (S106), and the update calculation process for the current cycle is completed.

[0100] On the other hand, if the ship speed range Δu is less than the range threshold (S103, NO), the update calculation unit 13 sets the regression equation to a zero-order equation (S107) and executes a subroutine process (S105).

[0101] 4.2.2 Subroutine Processing The operation of the subroutine process will be explained. Figure 6 is a flowchart showing the operation of the subroutine process.

[0102] As shown in Figure 6, first, the update calculation unit 13 calculates an initial value for the estimated value (S201) and determines whether the evaluation quantity has converged (S202). Here, the update calculation unit 13 makes this determination based on the following formula.

[0103]

number

[0104]

number

[0105]

number

[0106] If the evaluation quantity has not converged (S202, NO), the update calculation unit 13 calculates the regression coefficients of the order set in the update calculation process and the regression equation as an update function (S203), calculates the estimated value based on the calculated update function (S204), and then determines again whether the evaluation quantity has converged or not (S202).

[0107] On the other hand, if the evaluation quantity converges (S202, YES), the update calculation unit 13 terminates the subroutine processing.

[0108] 4.3 Nominal value updated The nominal values ​​of the hull parameters are used for setting values ​​and gain calculations of the control system, and are calculated from updated values ​​in equation (31), rather than using the identification values ​​obtained immediately beforehand. The updated values ​​are determined based on the corresponding ship speed.

[0109] The regression coefficient of the updated value is affected by the applicable range of ship speed.

[0110] If the ship speed range is narrow, the update value is:

[0111]

number

[0112] When the ship speed range is wide, the updated value is as shown in Figure 4.

[0113]

number

[0114] 5. Verification The effects of the automatic steering system for ships according to this embodiment will be verified by simulation.

[0115] 5.1 The effect of outlier removal Figure 7 shows the results of outlier removal. In Figure 7, (x,y)=(3,0) is considered an outlier. From Figure 7, the following can be seen.

[0116] • Outlier removal weights w O This is an outlier and is zero. • The estimated value is y h = y^= x

[0117] Therefore, the effectiveness of outlier removal was confirmed.

[0118] 5.2 Verification using identification data In the simulation for this verification, the identification dataset consists of 10 data points. The ship speed range is Δu = 1.6 set Since = 3kn, the regression coefficient is of the zeroth degree. However, the evaluation quantity J is of the zeroth degree here. reg0 and linear ones J reg1 We find the values ​​and adopt the smaller one. Note that we used η0 = 0.01 and η1 = 0.001.

[0119] 5.2.1 Evaluation Quantities and Their Weights Figure 8 shows the evaluation quantity and the weight of the evaluation quantity. In Figure 8, w J The difference can be as much as 10 times.

[0120] 5.2.2 Comparison of Conventional Method and Proposed Method We compare the conventional method and the proposed method. Here, the conventional method refers to the method of calculating the update value with a weight always set to 1, while the proposed method refers to the method in which the weight is varied according to this embodiment. Figure 9 is a table showing the update values ​​obtained using the conventional method and the proposed method, respectively. Figure 10 shows the turning force gain K as the update value obtained using the conventional method. r This figure shows the result. Figure 11 shows the turning force gain K as an updated value according to the embodiment. r This figure shows the result. Figure 12 shows the transverse flow gain K as the updated value using the conventional method. v This figure shows the result. Figure 13 shows the transverse flow gain K as the updated value according to the embodiment. v This figure shows the time constant T as the updated value using the conventional method.​r This figure shows the time constant T as the update value according to the embodiment. r This figure shows the time constant T as the updated value using the conventional method. r3 This figure shows the time constant T as the update value according to the embodiment. r3 This is a diagram.

[0121] The following can be seen from Figures 9 to 12. As shown in Figures 10, 12, 14, and 16, the weight w is 1 in the conventional method, but as shown in Figures 11, 13, 15, and 17, the weight w changes in the proposed method and can even be zero. The error err = w(P^-P) in the proposed method is less variable and more constant compared to the conventional method.

[0122] Furthermore, the following can be seen from Figure 9. • The updated values ​​obtained using the proposed method are approximately half of those obtained using the conventional method. The evaluation quantity J obtained by the proposed method is approximately 1 / 100th of the evaluation quantity J obtained by the conventional method.

[0123] 6. Summary As described above, the automatic steering system 1 for ships according to this embodiment estimates the updated values ​​of hull parameters in the identification system and selects regression coefficients according to the ship speed range by applying a robust estimation regression equation. The details of the study according to this embodiment are summarized below.

[0124] The identification system was explained, and statistical assumptions about the identification values ​​were established. The introduction of weights, outlier removal, and the calculation of regression coefficients were explained. • We explained how to handle ship speed, calculate the update function, and update the nominal value. • In the verification process, the effectiveness of the proposed method was confirmed through simulation.

[0125] The embodiments of the present invention are presented as examples and are not intended to limit the scope of the invention. These novel embodiments can be carried out in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims and their equivalents. [Explanation of Symbols]

[0126] 1. Automatic steering system for ships 12 Identification Calculation Unit 13 Update calculation section 22 hull

Claims

1. A ship's automatic steering system that outputs a command rudder angle using hull parameters based on the reference bearing and the heading, An identification calculation unit has an identification model that includes a hull model relating to the hull motion of a ship, which outputs model output data from predetermined input data obtained from the aforementioned automatic steering system for ships, and calculates an identification value by adjusting the hull parameters such that an evaluation quantity, which is the sum of squared errors between the model output data and the output data which are measured values ​​relating to the hull, is minimized during an identification value calculation period that includes at least the start time of the ship's change of course. An update calculation unit calculates coefficients for an update function, which is a regression equation relating to the ship's speed, based on an identification data sequence in which the calculated identification values ​​and the evaluation quantities corresponding to the identification values ​​are accumulated in a time series, and calculates updated values ​​for the ship's parameters using the update function to which the calculated coefficients are given. A ship's automatic steering system equipped with the following features.

2. The automatic steering device for ships according to claim 1, characterized in that the update calculation unit sets the order of the regression equation to 1 when the range of ship speeds corresponding to the identification data sequence is greater than or equal to a preset range threshold, and sets the order of the regression equation to 0 when the range of ship speeds corresponding to the identification data sequence is less than a preset range threshold.

3. The automatic steering device for ships according to claim 1, characterized in that the update calculation unit calculates the coefficients of the update function based on the weights of the evaluation quantities, which are the reciprocals of the normalized evaluation quantities.

4. The automatic steering device for ships according to claim 3, characterized in that the update calculation unit calculates the coefficients of the update function by excluding outliers of the identified values ​​in the identified data sequence based on the weights of the evaluation quantities.

Images