JP2008033505A - Method and device for retrieving data, and controller - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To speedily retrieve data accumulated in a data base. <P>SOLUTION: When a request point P1 being a new state vector is given, the difference between the element 1 of the request point P1 and the element 1 of each state vector in the data base is calculated to narrow the data to be state vectors in the prescribed range S1 of the difference. Regarding the narrowed state vectors, the difference between the element 2 of the request point P1 and the element 2 of each narrowed state vector is calculated to narrow the data to be state vectors in the prescribed range S2 of the difference. Regarding state vectors in a narrowed area S3, a distance (L1 norm) between each of the state vector and the request point P1 is calculated to select neighboring state vectors. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a data search method, a search device, and a control device using the same, which are suitable for searching database data at high speed.

  In a conventional control device, for example, a PID control device, auto-tuning is performed using a step response method or a limit cycle method, and control is performed by adjusting a PID control parameter. The system is identified and the PID control parameter is changed according to the identification result.

  However, in such a conventional example, particularly in a system with strong nonlinearity, there is a problem that reliability of identification is not sufficient and desired control characteristics may not be obtained.

Therefore, the applicant of the present application has already proposed a control device that can obtain more accurate control characteristics according to the system (see Patent Document 1).
JP 2005-339004 A

  In the proposed control device, the control system feature quantity and control parameters are stored in the database as information vectors, and when a new request point is given, an information vector close to the request point is searched from the database. Based on the retrieved information vector, a control parameter corresponding to the request point is generated, which is based on the so-called Just-In-Time (JIT) method or Model-on-Demand (MOD) method.

  The search for the information vector close to the request point is performed by calculating the distance between all the information vectors in the database and the request point, for example, the weighted distance or the Mahalanobis distance, and selecting the information vector within a predetermined distance, For this reason, there is a problem that the amount of calculation increases according to the scale of the database and the processing time becomes long.

  The present invention has been made in view of the above-described problems, and an object of the present invention is to provide a data search method, a data search device, and a control device using the same, which are designed to speed up data search.

  (1) The data retrieval method of the present invention is a data retrieval method for retrieving data in the vicinity of a certain state vector from a database in which state vector data is stored, and for each element constituting the state vector, data And a second step of searching for the neighboring data from the narrowed-down data.

  Here, the state vector refers to a feature amount represented by a vector, and can also be referred to as an information vector or a feature vector.

  To narrow down means to narrow the range of data to be searched.

  To narrow down the data for each element, for example, narrow down the data by the first element, narrow down the data by the first element, further narrow the data by the second element, and narrow the data by the second element. Further, the narrowed-down data may be further narrowed down, such as narrowing down by the third element, or all data in the database may be narrowed down individually for each element. May be.

  The narrowing may be performed based on at least one element of all elements constituting the state vector.

  According to the data search method of the present invention, the data in the database is narrowed down for each element constituting the state vector, and the nearby data is searched from the narrowed-down data, so the nearby data is searched from all the data in the database. Compared to the case, the time required for the search can be shortened.

  (2) In one embodiment of the data search method of the present invention, in the first step, a difference between an element of the certain state vector and an element of the state vector stored in the database is calculated, Narrow down the data by extracting the state vector whose calculated difference is within the predetermined range, and in the second step, calculate the distance between the certain state vector and the extracted state vector to search for the nearby data You may make it do.

  The distance refers to a vector distance. For example, a weighted distance or a Mahalanobis distance is preferable.

  According to this embodiment, the difference between the element of a certain state vector and the element of the state vector of the database is extracted by extracting data within a predetermined range, and the certain state is selected from the narrowed data. Since data in the vicinity of the vector is searched, the time required for the search can be shortened.

  (3) In the embodiment of the above (2), the first step includes a determination step for determining the predetermined range, and the determination step includes a maximum value of the elements of the state vector stored in the database. The minimum value may be searched for, and the predetermined range may be determined to be a range narrower than the difference between the searched maximum value and minimum value.

  The narrow range is preferably a range defined by a certain ratio with respect to the element range which is the difference between the maximum value and the minimum value of the element, but the ratio is the data stored in the database. It may be variable depending on the amount. The ratio may be the same ratio or a different ratio for each element.

  According to this embodiment, the range narrower than the difference between the maximum value and the minimum value of the state vector elements of the database is set as the predetermined range, and the difference between an element of a certain state vector and the element of the database state vector is the predetermined range. Whether or not data can be extracted can be determined depending on whether or not it is within the range.

  (4) In the embodiment of the above (2), the first step includes a determination step for determining the predetermined range, and the determination step is performed in advance with respect to the elements of the state vector stored in the database. The stored maximum value and minimum value may be updated, and the predetermined range may be determined to be a range narrower than the difference between the updated maximum value and minimum value.

  According to this embodiment, the maximum value and the minimum value stored in advance are updated, and the predetermined range is determined based on the difference between the updated maximum value and the minimum value, so that the state vector elements stored in the database are Compared with the configuration for searching for the maximum value and the minimum value, the time required for the search can be shortened.

  (5) In the embodiment of (2), the predetermined range may be a range centered on the element of the certain state vector.

  According to this embodiment, whether or not a difference between an element of a certain state vector and an element of the state vector stored in the database is within a predetermined range defined around the element of the certain state vector. Whether to extract the state vector can be determined.

  (6) In a preferred embodiment of the data search method of the present invention, in the first step, the data is narrowed down based on the first element among a plurality of elements constituting the state vector. The data may be narrowed down based on the element, and similarly, the narrowed data may be further narrowed down sequentially based on the next element.

  According to this embodiment, since the data narrowed down by a certain element is further narrowed down by the next element, all the data accumulated in the database are narrowed down individually for each element. Search processing can be performed at high speed.

  (7) A data search device according to the present invention is a data search device for searching data in the vicinity of a certain state vector from a database in which state vector data is stored, and for each element constituting the state vector, data And means for searching for the neighboring data from the narrowed-down data.

  According to the data search device of the present invention, the data in the database is narrowed down for each element constituting the state vector, and the nearby data is searched from the narrowed-down data, so the nearby data is searched from all the data in the database. Compared to the case, the time required for the search can be shortened.

  (8) In the control device of the present invention, a control unit that controls a control target according to a control parameter, a vector that includes a control system feature value is accumulated as a state vector, and a control system characteristic parameter that includes the control parameter is Adjusting means for adjusting a control parameter of the control means based on data stored in the database, and the adjusting means is provided with a new state vector, whereby a state vector in the vicinity of the new state vector is provided. A data search unit that searches the database from the database, and a characteristic parameter generation unit that generates a characteristic parameter corresponding to the new state vector based on the searched state vector, and the characteristic generated by the characteristic parameter generation unit The parameter is stored in the database, and the data search unit For each element constituting the state vector, the difference between the new state vector element and the state vector element stored in the database is calculated, and a state vector in which the calculated difference is within a predetermined range is extracted. The distance between the extracted state vector and the new state vector is calculated to search for nearby state vectors.

  Here, the characteristic amount of the control system refers to an amount representing the characteristic of the control system, and refers to an amount representing the characteristic of the control target, the control device, or the environment in the control system including the control target, the control device, and the environment. For example, the control amount, the operation amount, the control error, the ambient temperature, the magnitude of the disturbance, and the like.

  The feature amount preferably includes at least one of the control amount time-series data, or preferably includes at least one of the operation amount time-series data and the set value.

  Furthermore, it is preferable that the set value is an output of a reference model. Here, the reference model refers to a model showing desired response characteristics, for example, a model having characteristics of a transfer function of a first-order lag or a second-order lag.

  Further, the characteristic parameter of the control system refers to a parameter that represents the characteristic of the control system, and in the control system configured by the control target, the control device, and the environment, refers to a parameter that represents the characteristic of the control target, the control device, or the environment. For example, it refers to a coefficient of a transfer function to be controlled, a time constant of the control target, a dead time, a steady gain, or a control parameter.

  The characteristic parameters of the control system include control parameters of the control means, for example, PID control parameters.

  It is preferable that the control system characteristic amount and the control system characteristic parameter are stored in the database in association with each other.

  Preferably, the characteristic parameter generation unit configures the characteristic parameter as a local model using a weighted linear average method or a weighted local regression method based on the state vector of the database.

  The adjustment means searches for a state vector in the vicinity of the new state vector from the database when a new state vector as a request point is given, and responds to the new state vector based on the searched state vector. The characteristic parameter to be generated is generated.

  The state vector in the vicinity of the new state vector refers to a state vector whose vector is close to the new state vector, or a state vector similar to the new state vector. This distance is preferably a weighted distance or a Mahalanobis distance.

  According to the control device of the present invention, the feature quantity and characteristic parameter of the control system are accumulated in the database, and when a new state vector is given, a new state vector is generated as a local model based on the accumulated state vector of the database. Control parameters are adjusted based on the so-called Just-In-Time (JIT) method or Model-on-Demand (MOD) method, which generates characteristic parameters corresponding to Highly accurate control characteristics can be obtained.

  In addition, the data search unit calculates a difference between the new state vector element and the state vector element stored in the database for each element constituting the state vector, and the calculated difference falls within a predetermined range. Since an existing state vector is extracted and a distance between the extracted state vector and the new state vector is calculated to search for nearby state vectors, all the state vectors stored in the database and the new state vector Compared to the configuration in which the distance is calculated and the nearby state vector is searched, the search process can be speeded up, and this enables high-precision control.

  (9) In one embodiment of the control device of the present invention, the data search unit searches for the maximum value and the minimum value of the elements of the state vector stored in the database every time a new state vector is given. The predetermined range may be determined to be narrower than a difference between the searched maximum value and minimum value.

  According to this embodiment, a range narrower than the difference between the maximum value and the minimum value of the state vector elements of the database is set as the predetermined range, and the difference between the new state vector element and the database state vector element is the predetermined range. Whether or not data can be extracted can be determined depending on whether or not it is within the range.

  (10) In another embodiment of the control device of the present invention, each time a new state vector is given, the data search unit stores a maximum value stored in advance for the elements of the state vector stored in the database, and The minimum value is updated, and the predetermined range is determined to be a range narrower than the difference between the updated maximum value and minimum value.

  According to this embodiment, every time a new state vector is given, the maximum value and minimum value stored in advance are updated, and the predetermined range is determined based on the difference between the updated maximum value and minimum value. Each time a state vector is given, it is not necessary to search for the maximum value and the minimum value of the elements of the state vector stored in the database, and the time required for the search can be shortened.

  (11) In the embodiments of (9) and (10), the data search unit may set the predetermined range as a range obtained by multiplying a difference between the maximum value and the minimum value by a set value.

  This set value may be set by operating a setting unit of the control device, or may be set by communication from the outside.

  The set value is preferably a ratio to the difference between the maximum value and the minimum value of the elements.

  According to this embodiment, a range narrower than the difference between the maximum value and the minimum value of the state vector elements of the database is set as the predetermined range, and the difference between the new state vector element and the database state vector element is the predetermined range. Whether or not data can be extracted can be determined depending on whether or not it is within the range.

  (12) In another embodiment of the control device of the present invention, the predetermined range may be a range centered on the element of the new state vector.

  According to this embodiment, whether or not the difference between the element of the new state vector and the element of the state vector stored in the database is within a predetermined range defined around the element of the new state vector Whether to extract the state vector is determined.

  (13) In a preferred embodiment of the control device of the present invention, the data search unit extracts a state vector based on a first element among a plurality of elements constituting the state vector, and for the extracted state vector, The state vector may be extracted based on the next element, and similarly, the extracted state vector may be sequentially extracted based on the next element.

  According to this embodiment, since the data narrowed down by a certain element is further narrowed down by the next element, all the data stored in the database is narrowed down individually for each element. The search process becomes faster.

  According to the present invention, the data in the database is narrowed down for each element constituting the state vector, and nearby data is searched from the narrowed-down data. Therefore, when searching for nearby data from all the data in the database, In comparison, the time required for the search can be shortened.

  Also, state vectors and characteristic parameters including control system features are stored in the database. When a new state vector is given, a nearby state vector is searched from the stored database, and based on the searched state vector Since a characteristic parameter corresponding to a new state vector is generated as a local model, control parameters are adjusted based on the so-called Just-In-Time (JIT) method or Model-on-Demand (MOD) method. Even in a highly reliable system, highly accurate control characteristics can be obtained.

  Moreover, the search for nearby state vectors is performed by narrowing down the state vectors stored in the database for each element constituting the state vector and calculating the distance from the new state vector for the narrowed state vector. Compared to the conventional example in which the distances between all the state vectors stored in the database and the new state vector are calculated, it is possible to increase the speed of the search process, thereby enabling highly accurate control.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram of a main part of a temperature regulator as a control device according to one embodiment of the present invention.

  The temperature controller 1 of this embodiment performs temperature control of the control target 2 such as a heat treatment apparatus, and is disposed on the control target 2 and a set temperature r set from a host device or a setting unit (not shown). An adjustment for adjusting a PID control parameter in the PID controller 3 is provided with a PID controller 3 as a control means for calculating and calculating the manipulated variable u based on a deviation e from a detected temperature y from a temperature sensor (not shown). Means 4 are provided.

  In the temperature controller 1 of this embodiment, the following is performed in order to obtain highly accurate control characteristics.

  That is, each time new data is obtained, it is stored in the database 6 and, if necessary, a local model is created by taking out similarities to the requested points from a large number of databases 6 accumulated in the past as neighborhoods. The PID control parameter of the PID controller 3 is adjusted based on a so-called Just-In-Time (JIT) method.

  First, the JIT modeling method will be described.

  Consider a discrete-time nonlinear system expressed by the following equation.

上式は、ＮＡＲＭＡＸ（Ｎonlinear ＡＲＭＡＸ）モデルと呼ばれ、ダイナミクスを持つ大域的な非線形システムを表すモデル表現として広く使用されている。ここで、ｙ（ｔ）はシステム出力、ｆ（・）は非線形関数、ｅ（ｔ）は観測雑音、φ（ｔ）はシステムの時刻ｔでの状態を表しており、状態ベクトルと呼ぶこととする。状態ベクトルφ（ｔ）は次式で定義される。

The above equation is called a NARMAX (Nonlinear ARMAX) model, and is widely used as a model expression representing a global nonlinear system having dynamics. Here, y (t) is a system output, f (•) is a nonlinear function, e (t) is observation noise, φ (t) is a state of the system at time t, and is called a state vector. To do. The state vector φ (t) is defined by the following equation.

ここで、ｕ（ｔ）はシステム入力、ｄはむだ時間、ｎ_ｙ，ｎ_ｕはそれぞれ出力と入力の次数である。ＪＩＴモデリングでは、（２）式の状態ベクトルの形式でデータベースへの蓄積が行われる。また、出力ｙ（ｔ）の予測値を得るために必要な状態ベクトルφ（ｔ）を要求点ｑと呼ぶ。ＪＩＴモデリング法とは、この要求点に類似した状態ベクトルをデータベースから近傍として抽出し、局所モデルを構成する方法である。

Here, u (t) is a system input, d is a dead time, and n _y and n _u are output and input orders, respectively. In JIT modeling, data is stored in a database in the state vector format of equation (2). Further, the state vector φ (t) necessary for obtaining the predicted value of the output y (t) is called a request point q. The JIT modeling method is a method of extracting a state vector similar to this request point as a neighborhood from a database and constructing a local model.

  In this embodiment, the PID control parameter of the PID controller 3 in the block diagram of FIG. 1 is adjusted based on the JIT method as described above.

  Here, a speed type I-PD control law as shown below is applied.

但し、ｅ(ｔ)は制御誤差信号であり、ｒ(ｔ)を目標値とすると、以下で定義される。

However, e (t) is a control error signal, and is defined below, where r (t) is a target value.

また、ｋｃ、Ｔ_Ｉ、およびＴ_Ｄはそれぞれ比例ゲイン、積分時間、微分時間を、Ｔｓはサンプリング時間を表しており、(４)式では、簡単のために、Ｋｐ＝ｋｃ、Ｋ_Ｉ＝ｋｃＴｓ／Ｔ_Ｉ、Ｋ_Ｄ＝ｋｃＴ_Ｄ／Ｔｓと置き換えている。さらに、Δ(：＝１−ｚ^−１)は差分演算子を表している。

Further, kc, T _I , and T _D represent proportional gain, integration time, and derivative time, respectively, and Ts represents sampling time. In the equation (4), for simplicity, Kp = kc, K _I = kcTs / T _I , K _D = kcT _D / Ts Further, Δ (: = 1−z ⁻¹ ) represents a difference operator.

  The basic idea of this method is to obtain PID control parameters as an inverse model such as the following equation by the JIT method.

但し、ｙ_ｒは参照モデルの出力を示しており、次式のように設計される。

However, y _r denotes the output of the reference model is designed as follows.

上式に含まれるＴ(ｚ^−１)については、Ｔ．Ｙamamoto and Ｓ．Ｌ．Ｓhah： Ｄesign and Ｅxperimental Ｅvaluation of ａ Ｍultivariable Ｓelf-Ｔuning ＰＩＤ Ｃontroller,Ｐroc.of ＩＥＥＥ Ｃonference on Ｃontrol Ａpplications,Ｔrieste,pp.1230-1234(1998)に基づいて設計する。

Regarding T (z ⁻¹ ) included in the above equation, T.W. Yamamoto and S. L. Shah: Designed based on Design and Experimental Evaluation of a Multi-variable Self-Tuning PID Controller, Proc. Of IEEE Conference on Control Applications, Trieste, pp. 1230-1234 (1998).

The JIT-PID controller algorithm is shown below.
[STEP 1] Creation of initial database In the JIT method, if there is no past accumulated data, it is impossible to identify in principle. Create
[STEP 2] Distance calculation, neighborhood selection A state vector necessary for obtaining a predicted value of the PID parameter K (t) at time t is called a request point q. The distance between the request point q and the state vector stored in the database is obtained by the weighted L1 norm of the following equation.

この距離が小さいものからｋ個のベクトルを近傍として選択する。
［ＳＴＥＰ３］ 局所モデルの構成
続いて、ＳＴＥＰ２において選択された近傍に対して、以下で示される、重みつき局所線形平均法（ＬＷＡ）により局所モデルを構成する。

K vectors having the smallest distance are selected as neighbors.
[STEP 3] Configuration of Local Model Next, a local model is configured for the neighborhood selected in STEP 2 by the weighted local linear average method (LWA) shown below.

ここで、ｗｉは、近傍の第ｉサンプルに対応する重みであり、次式で与える。

Here, wi is a weight corresponding to the i-th sample in the vicinity, and is given by the following equation.

［ＳＴＥＰ４］ データ修正
ＳＴＥＰ３において、局所モデルとして得られるＰＩＤ制御パラメータＫ^ｏｌｄは、制御誤差（ε：＝ｙ_ｒ−ｙ）を考慮していないため、ＳＴＥＰ２、ＳＴＥＰ３を繰り返し行い、ＰＩＤ制御パラメータを逐次抽出したとしても、システムの状態に応じてＰＩＤ制御パラメータが適切に調整されない恐れがある。そこで、ＳＴＥＰ２で得られたＰＩＤ制御パラメータＫ^ｏｌｄに対して制御誤差εにより修正を行ない、その修正されたデータＫ^ｎｅｗをデータベースに蓄えるものとする。

In [STEP4] data correction STEP3, PID control parameters ^{K old} obtained as the local model, the control error _{(ε: =} y r -y) because it does not consider, STEP2, repeats the STEP3, sequential PID control parameters Even if it is extracted, the PID control parameter may not be adjusted appropriately according to the state of the system. Accordingly, it is assumed that the PID control parameter K ^old obtained in STEP 2 is corrected by the control error ε, and the corrected data K ^new is stored in the database.

  The correction method is given by the following equation.

ここで、ηは学習係数、Ｊは以下で定義される誤差の評価規範を表している。

Here, η represents a learning coefficient, and J represents an error evaluation standard defined below.

また、（１３）式、右辺第２項のそれぞれの偏微分は、以下のように展開される。

Further, the partial differentials of the expression (13) and the second term on the right side are expanded as follows.

この実施の形態では、上述のようにＪＩＴ法に基づいて、ＰＩＤ制御器３のＰＩＤ制御パラメータを調整するために、調整手段４は、参照モデル５と、データベース６と、ＰＩＤチューニング部７とを備えている。

In this embodiment, in order to adjust the PID control parameter of the PID controller 3 based on the JIT method as described above, the adjusting unit 4 includes the reference model 5, the database 6, and the PID tuning unit 7. I have.

  In this embodiment, as shown in FIG. 2, the database 6 stores data consisting of a state vector, which is a vector including a control system feature quantity, and a PID control parameter corresponding to this state vector. The That is, a set of PID control parameters corresponds to a set of state vectors.

In this embodiment, as shown in FIG. 3, the state vector is composed of the output yr of the reference model 5 and the time series data of the detected temperature y of the controlled object 2 that is the controlled variable, and the PID control parameter is Kp = kc, K _I = kcTs / T _I , and K _D = kcT _D / Ts.

  Yr (t + 1), y (t), y (t−1),... Y (t−n) shown in FIG. 3 are each element constituting the state vector. For example, in the state vector of data number 1, 100, 98, 96,... 90 are each element constituting the state vector.

  As shown in the functional block diagram of FIG. 4, the PID tuning unit 7 of this embodiment includes a data search unit 8 that searches for nearby data, a local model generation unit 9 as a PID control parameter generation unit, and a data The correction part 10 is provided.

  In the data search unit 8, every time the above-described request point is given, that is, every time a state vector at the latest time t is given, this state vector and the database 6 narrowed down as described later are accumulated. The distance from the state vector is obtained by the weighted L1 norm of the above equation (10), and k state vectors having the smallest distance are selected as the neighborhood.

The local model generation unit 9 obtains the PID control parameter K ^old by the weighted local linear average method for the selected nearby state vector. That is, a weight is determined according to the above equation (12) using the k neighboring state vectors selected, and this weight is determined as a PID control parameter corresponding to the k state vectors according to the above equation (11). Is multiplied (by weighted average) to obtain the PID control parameter K ^old .

The PID control parameter K ^old generated by the local model generation unit 9 is set in the PID controller 3.

The data correction unit 10 corrects the PID control parameter K ^old generated by the local model generation unit 9 in such a direction that the control error is reduced as shown in the above-described equations (13) to (17). The corrected PID control parameter K ^new is stored in the database 6 in association with the state vector.

  In this data correction unit 10, in order to correct based on the control error, the output yr of the reference model 5 and the detected temperature y that is a control amount are given.

  In this embodiment, the PID controller 3 and the adjusting means 4 are constituted by, for example, a microcomputer, and by executing a program stored in the ROM of the microcomputer, the PID control parameters are adjusted based on the JIT method. The above-mentioned database 6 is configured on a RAM of a microcomputer, for example.

  In this embodiment, the amount of data that can be stored in the database 6 is limited, and when the control object 2 changes with time, it is preferable to delete the old data. Is trying to delete from the old data.

  As described above, numerical examples are shown below to evaluate the effectiveness of the temperature controller 1 for adjusting the PID control parameter of the PID controller 3 based on the JIT method.

  Here, a polystyrene polymerization reactor is considered as a control target. The relational expression between the jacket temperature u (t) and the reaction temperature y (t) in the reactor is given by the following equation (Nakanishi, Hanakuma: Process Control Basics and Practice, Asakura Publishing, (1992) ).

ここで、Ｅａ＝２４０、Ｒ＝０．０１９８６である。このシステムは、目標値が高くなる(約７５度以上)につれ非線形性が著しく強くなるという特徴を持っている。

Here, Ea = 240 and R = 0.01986. This system has a feature that the nonlinearity becomes remarkably stronger as the target value becomes higher (about 75 degrees or more).

  A PID control parameter adjustment method based on the JIT method is applied to this system.

Here, the design polynomial T (z ⁻¹ ) included in the reference model is designed as follows.

ここで、Ｔ(ｚ^−１)は、上述の文献Ｔ．Ｙamamoto and Ｓ．Ｌ．Ｓhah： Ｄesign and Ｅxperimental Ｅvaluation of ａ Ｍultivariable Ｓelf-Ｔuning ＰＩＤ Ｃontroller,Ｐroc.of ＩＥＥＥ Ｃonference on Ｃontrol Ａpplications,Ｔrieste,pp.1230-1234(1998)を参考にして設計した。

Here, T (z ⁻¹ ) is the above-mentioned document T.I. Yamamoto and S. L. Shah: Designed with reference to Design and Experimental Evaluation of a Multivariable Self-Tuning PID Controller, Proc. Of IEEE Conference on Control Applications, Trieste, pp. 1230-1234 (1998).

  FIG. 5 shows the control result at this time, and FIG. 6 shows the temporal change of the PID control parameter.

  As shown in these figures, it can be seen that the PID control parameters are appropriately adjusted online according to the nonlinearity of the system, and a good control result is obtained.

  Conventionally, every time a request point is given, that is, every time a state vector at the latest time t is given, the distance between this state vector and all the state vectors stored in the database 6 is expressed as ( Obtained by the weighted L1 norm of equation (10), k state vectors having the smallest distance are selected as neighbors.

  FIG. 7 is a flowchart showing the conventional neighborhood selection procedure.

  First, for all data in the database 6, the distance from the new state vector that is the request point is calculated and stored (step n1).

  Specifically, for the elements constituting the state vector, first, the element I = 1 is set (step n1-1), and the maximum value and the minimum value of the element I are set for all data (all state vectors) in the database 6. Search and store as maxφ (m) I, minφ (m) l (step n1-2), determine whether I has reached the total number of elements (step n1-3), and if not, Similar processing is performed for the next element I + 1 (step n1-4) (step n1-2). This is because the search for the maximum value and the minimum value for each element in step n1-2 is used for the calculation of the distance by the above equation (10).

  When the search for the maximum value and the minimum value is completed for all the elements, the process proceeds to step n1-5, where the data j = 1 is set, and the distance between the request point and the data j is calculated by the above equation (10). d (q, φ (j)) is stored (step n1-6), and it is determined whether or not the distance calculation and storage have been completed for all data (step n1-7). Move.

  In step n2, all the data in the database 6 are sorted in order of distance from the request point, and k items are selected from the closest to the request point as the neighborhood (step n3).

  As described above, the selection of the neighborhood has a problem that the calculation amount increases and the processing time becomes long in order to calculate the distance between all the data in the database 6 and the request point.

  FIGS. 8 and 9 are diagrams for explaining this conventional neighborhood selection, and show an example in which the state vector is composed of two elements, element 1 and element 2, for the sake of simplicity. In these figures, as a request point, an example in which the value of element 1 is “0.45” and the value of element 2 is “0.55” is shown.

  When a request point is given, a distance between the request point and all data in the database 6, for example, a distance typically indicated by an arrow in FIG. 8 is calculated, and a state close to the request point as shown in FIG. Several points of the vector will be selected.

  On the other hand, in the data search method of this embodiment, instead of calculating the distance to the requested point for all data in the database 6, the data is narrowed down for each element of the state vector, and the narrowed data is compared with the requested point. By calculating the distance, the processing speed is increased.

  FIGS. 10 to 12 are diagrams corresponding to FIGS. 8 and 9 for explaining the neighborhood selection of this embodiment, and the state vector is composed of two elements, element 1 and element 2, as in the conventional example. An example is shown in which the value of element 1 is “0.45” and the value of element 2 is “0.55” as the request point P1.

  In this embodiment, the data is narrowed down for each element. When the request point P1 is given, as shown in FIG. 10, first, based on the element 1, the element of the request point P1 is centered on the request point P1. The data is narrowed down to the data in the region S1 surrounded by a rectangle whose difference from 1 is within a predetermined range. In this example, the predetermined range is a range with a width of ± 0.15 centered on the required point with respect to the difference between the minimum value and the maximum value of the elements before narrowing down. In this way, the difference from the element of the request point P1 is narrowed down to the data of the element within the predetermined range, that is, the element approximated to the element of the request point P1.

  Next, with respect to the narrowed-down data, as shown in FIG. 11, the difference between the request point P2 and the element 2 around the request point P1 is within a predetermined range based on the element 2 as in the case of the element 1. The data is narrowed down to the data of the area S2 surrounded by the rectangle inside, and therefore the data is narrowed down to the data of the rectangular area S3 where the rectangular area S1 based on the element 1 and the rectangular area S2 based on the element 2 overlap. become.

  With respect to the data thus narrowed down, the distance from the request point P1 is calculated as shown by the arrow in FIG. 11, and as shown in FIG. 12, several points of the state vector close to the request point P1 are calculated. Will be selected.

  FIG. 13 is a diagram illustrating a part of a numerical example of a data narrowing simulation when the number of elements is two.

  In the figure, the required points are the value of element 1 is “0.45”, the value of element 2 is “0.55”, the values of element 1 and element 2 for each data, Difference between element 1 and element 1 of request point, availability of data extraction based on difference of element 1, difference between element 2 and element 2 of request point for each data, data difference based on difference of element 2 The extraction possibility and the value of the distance (L1 norm) from the request point for the extracted data are shown, respectively. Regarding the extraction possibility, the extracted data is “1” and the non-extracted data is “0”. Show.

  For each data, the difference between element 1 of each data and element 1 of the requested point is calculated, and the absolute value of the difference is centered on the value “0.45” of element 1 of the requested point, for example ± 0 .. 15 is determined depending on whether or not it is within a predetermined range of the width, and the data is narrowed down.

  For this narrowed-down data, the difference between the element 2 of each data and the element 2 of the request point is calculated, and the absolute value of the difference is centered on the value “0.55” of the element 2 of the request point. , Whether or not extraction is possible is determined depending on whether or not it is within a predetermined range of ± 0.15 width (extraction width), and data is narrowed down.

  The distance (L1 norm) from the request point is calculated for the narrowed data.

  Note that FIG. 13 shows the difference for element 2 and the possibility of extraction based on the difference for all data regardless of the possibility of extraction of data based on element 1.

  FIG. 14 shows a conventional example corresponding to FIG. As shown in FIG. 14, in the conventional example, the distance (L1 norm) is calculated for all data. In FIG. 14, the difference between the elements 1 and 2 is also shown.

  FIG. 13 shows the nearest distance (shortest distance), the next closest distance (second place), and the next closest distance (third place) among the k pieces of data retrieved by this conventional example. The search results according to the preferred embodiment are in good agreement.

  FIG. 15 is a flowchart showing the neighborhood selection procedure of this embodiment.

  First, a narrowing ratio p (0 <p <1), which is a ratio for narrowing elements, is set (step n1). This narrowing rate p is for defining a predetermined range centered on the request point, and is the difference between the minimum value and the maximum value of the elements for all data before narrowing down, that is, the ratio to the maximum width, As described above, when the width of ± 0.15 centered on the required point is within a predetermined range, this narrowing rate is 0.3.

Next, the minimum number of extracted data kmin is set (step n2). This minimum number of extracted data kmin is:
It is set so that k ≦ kmin <the number of data in the database.

  Next, the element I is set to 1 (step n3), and the data range for the element I is narrowed down (step n4).

  Specifically, the maximum value and the minimum value of element I are searched for all the data before narrowing down and stored as maxφ (m) I and minφ (m) I, respectively (step n4-1). Next, with respect to all the data after narrowing down, in the case of the first element that has not been narrowed down, with respect to all data in the database, a certain width around the requested point element, that is, the narrowing rate p * (maxφ (m ) Extract and narrow down data within a predetermined range of I-minφ (m) l) (step n4-2). Thus, the data narrowing based on one element is completed.

  Next, the process proceeds to step n5, where it is determined whether or not the number of extracted data is less than or equal to the minimum extracted data number kmin. When the number is less than the minimum extracted data number kmin, the process proceeds to step n6 and the minimum extracted data number is reached. When it is not less than kmin, it is determined whether or not the element I has reached the total number of elements (step n9). When the total number of elements has been reached, the process proceeds to step n6, and when the total number of elements has not been reached, The same narrowing is performed for the next element I + 1 (step n10) (step n4).

  In step n6, the distance to the request point is calculated and stored for the narrowed data, that is, for the extracted data. Specifically, the extracted data j = 1 is set (step n6-1), and the distance between the requested point and the extracted data j is calculated by the above-described equation (10) and d (q, φ (j)) (Step n6-2), and it is determined whether or not the calculation and storage of the distance have been completed for all the extracted data (step n6-3). When completed, the process proceeds to step n7, and when not completed Returning to step n6-2, the distance to the requested point is calculated and stored for the next data j + 1.

  In step n7, all the extracted data are sorted in the order of the distance from the request point, and k pieces are selected as the neighborhood from the closest distance to the request point (step n8).

  In this way, the neighborhood search is not performed by calculating the distance for all data in the database as in the conventional case, but is performed on the data narrowed down for each element of the state vector, so that the calculation amount is reduced and the processing speed is increased. Can be planned.

FIG. 16 is a numerical example showing a comparison result of the processing time between Example 1 according to the embodiment and the conventional example. In FIG. 16, as the common condition, the number of data is 10,000 and the number of elements constituting the state vector. 4, addition / subtraction time is 1.00 × 10 ⁻⁶ seconds / time, division time is 2.00 × 10 ⁻⁶ seconds / time, multiplication time is 1.00 × 10 ⁻⁶ seconds / time, comparison operation time is 1 0.000 × 10 ⁻⁶ seconds / time, which indicates the number of operations required for each process, the time required for the operation, and the total calculation time. In the first embodiment, the narrowing rate is set to 0.3.

In the conventional example shown in FIG. 16A, searching for the maximum value and minimum value used for calculating the distance, calculating the numerator and denominator in the distance calculation formula, and performing a comparison operation for sorting based on the calculated distance are required. And the total calculation time is 2.73 × 10 ⁻¹ seconds.

  On the other hand, in the first embodiment shown in FIG. 16B, first, for the element 1, the maximum value and the minimum value used for the calculation of the distance and the setting of the predetermined range based on the narrowing rate are searched, and the element 1 is used. An operation for narrowing down the data is necessary, and for the narrowed-down data, an operation for sequentially performing the narrowing down based on the elements 2, 3, and 4 is necessary.

  In the first embodiment, since the data is sequentially narrowed down for each element, it can be seen that the time required for the calculation becomes shorter as the calculation proceeds with the element 2, the element 3, and the element 4.

  After the calculation for each element is completed and the data is narrowed down, the numerator and denominator calculation in the distance calculation formula, the comparison calculation for sorting based on the calculated distance, and the like are performed as in the conventional case.

The total calculation time of Example 1 is 5.88 × 10 ⁻² seconds, and it can be understood that the calculation time is 22% of the conventional example.

  In the above-described embodiment, the maximum value and the minimum value are searched for each element every time a request point is given. However, as another embodiment of the present invention, as an initial process, the maximum value for each element is previously stored. When the data is added, updated, or deleted, for example, when a request point is given, the element of the request point and the maximum value and minimum value of the corresponding element are stored. And the maximum value and the minimum value may be sequentially updated so that the search for the maximum value and the minimum value may be omitted.

FIG. 17 shows the calculation time of the second embodiment in which the search for the maximum value and the minimum value performed each time a request point is given is omitted by previously storing and updating the maximum value and the minimum value. FIG. 16B is a diagram corresponding to FIG. 16B, and the total calculation time is 3.05 × 10 ⁻² seconds, which means that the calculation time of 11% of the conventional example of FIG. Can be achieved.

(Other embodiments)
As another embodiment of the present invention, the reference model may be omitted, and the set value r may be used as the state vector instead of the output of the reference model 5.

  In each of the above-described embodiments, description has been made by applying to PID control. However, the present invention is not limited to PID control, but can be applied to adjustment of other control parameters such as GPC (Generalized Predictive Control), for example. Of course.

  In the above-described embodiment, the present invention has been described by applying to a temperature controller that performs temperature control. However, the present invention is not limited to a temperature controller, and various pressures, flow rates, speeds, liquid levels, and the like to be controlled The present invention can be applied to a control device that controls a physical state.

  The data search method or data search apparatus of the present invention is not limited to the control device as in the above-described embodiment. For example, the search of image data, the search of data of various properties close to the customer's required specifications, and other general It can be applied to the search of database data.

  The present invention is useful for database data retrieval and the like.

It is a block diagram of the temperature regulator which concerns on one embodiment of this invention. It is a figure which shows the structure of the data accumulate | stored in the database of FIG. It is a figure which shows the specific example of the data of FIG. FIG. 2 is a functional block diagram of a PID tuning unit in FIG. 1. It is a figure which shows the time change of reaction temperature. It is a figure which shows the time change of a PID control parameter. It is a flowchart for demonstrating the procedure of the neighborhood selection of a prior art example. It is a figure for demonstrating the conventional neighborhood selection about the state vector comprised from two elements. It is a figure for demonstrating the conventional neighborhood selection about the state vector comprised from two elements. It is a figure for demonstrating the conventional neighborhood selection about the state vector comprised from two elements. It is a figure for demonstrating the conventional neighborhood selection about the state vector comprised from two elements. It is a figure for demonstrating the conventional neighborhood selection about the state vector comprised from two elements. It is a figure which shows the numerical example of the simulation of data narrowing down. It is a figure corresponding to FIG. 13 of a prior art example. It is a flowchart which shows the procedure of the vicinity selection of embodiment. It is a figure which shows the numerical example of the data search time of an Example and a prior art example. It is a figure which shows the numerical example of the data search time of another Example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Temperature controller 2 Control object 3 PID controller 4 Adjustment means 6 Database 8 Data search part

Claims (13)

A data search method for searching data in the vicinity of a certain state vector from a database in which state vector data is accumulated,
A first step of narrowing down data for each element constituting the state vector;
A second step of searching for the neighboring data from the narrowed-down data;
A data search method comprising: In the first step, a difference between the element of the certain state vector and an element of the state vector stored in the database is calculated, and a state vector in which the calculated difference is within a predetermined range is extracted to obtain data Refine
The data search method according to claim 1, wherein in the second step, a nearby data is searched by calculating a distance between the certain state vector and the extracted state vector.   The first step includes a determination step of determining the predetermined range, and the determination step searches for a maximum value and a minimum value of state vector elements stored in the database, and searches the predetermined range. The data search method according to claim 2, wherein a range narrower than a difference between the maximum value and the minimum value is determined.   The first step includes a determination step of determining the predetermined range, and the determination step updates a maximum value and a minimum value stored in advance for elements of the state vector stored in the database, and The data search method according to claim 2, wherein the range is determined to be narrower than a difference between the updated maximum value and minimum value.   The data search method according to claim 2, wherein the predetermined range is a range centered on the element of the certain state vector.   In the first step, the data is narrowed down based on the first element among the plurality of elements constituting the state vector, and the narrowed data is narrowed down based on the next element. Furthermore, the data search method of any one of Claims 1-5 which narrows down data sequentially based on the following element. A data search device for searching data in the vicinity of a certain state vector from a database in which state vector data is accumulated,
Means for narrowing the data for each element constituting the state vector;
Means for searching for data in the vicinity from the narrowed-down data;
A data search apparatus comprising: Based on the data of the database in which the control means for controlling the controlled object according to the control parameter and the vector including the control system characteristic amount are accumulated as the state vector, and the control system characteristic parameter including the control parameter is accumulated, Adjusting means for adjusting a control parameter of the control means,
The adjusting means is provided with a new state vector, whereby a data search unit for searching a state vector in the vicinity of the new state vector from the database, and the new state vector based on the searched state vector A characteristic parameter generating unit that generates a characteristic parameter corresponding to the characteristic parameter, and storing the characteristic parameter generated by the characteristic parameter generating unit in the database,
The data search unit calculates a difference between the element of the new state vector and the element of the state vector stored in the database for each element constituting the state vector, and the calculated difference is within a predetermined range. A control device, wherein a state vector is extracted, a distance between the extracted state vector and the new state vector is calculated, and a nearby state vector is searched.   The data search unit searches for a maximum value and a minimum value of elements of the state vector stored in the database each time a new state vector is given, and searches the predetermined range for the maximum value and the minimum value. The control device according to claim 8, wherein the control device is determined to be narrower than a difference between the control device and the controller.   The data search unit updates the maximum and minimum values stored in advance for the elements of the state vector stored in the database each time a new state vector is given, and updates the predetermined range to the maximum The control device according to claim 8, wherein the control device determines a range narrower than a difference between the value and the minimum value.   The control device according to claim 9 or 10, wherein the data search unit sets the predetermined range as a range obtained by multiplying a difference between the maximum value and the minimum value by a set value.   The control device according to claim 8, wherein the predetermined range is a range centered on the element of the new state vector.   The data search unit extracts a state vector based on a first element among a plurality of elements constituting the state vector, extracts a state vector based on the next element for the extracted state vector, and similarly, The control device according to any one of claims 8 to 12, wherein the extracted state vectors are further sequentially extracted based on the following elements.

Images