JP3776329B2 - Radar identification device - Google Patents

Description

【００２２】
ステップＳＴ３においては、入力されたレーダ波データに係る全てのパルスデータより得られたＴ個の局所周期成分（Ｘ_１〜Ｘ_Ｔ）それぞれについて、識別子を参照することなく所定数（ｎ個）だけ設定された局所パターンにおいていずれの局所パターンに帰属するかを決定する教師なし分類を実施する。図３は、局所周期成分に係る教師なし分類動作を示すフローチャートである。まず、それぞれが正規分布として与えられる複数の局所パターン全体を表現する混合正規分布を規定する各パラメータを以下に示す期待値最大化法で求めるための初期値を設定する（サブステップＳＴ１１）。例えば、レーダ波毎に与えられるＴ個の局所周期成分Ｘ_１〜Ｘ_Ｔからｎ個の互いに異なるものを適宜選択してそれぞれ各局所パターンについての平均の初期値とし、Ｘ_１〜Ｘ_Ｔから分散を算出してｎ個の局所パターンに共通な分散の初期値とし、混合比の初期値をｎ個の局所パターンそれぞれに共通に１／ｎとする。各局所パターンについて平均、分散、混合比の組（μ_ｉ，σ_ｉ，α_ｉ）（以下、分類パラメータと称する）が与えられれば、局所周期成分（Ｘ_１〜Ｘ_Ｔ）のそれぞれについて、各局所パターンにおいての確率密度ｐ_１〜ｐ_ｎを以下の式（２）に基づいて算出する。
ｐ_ｉ＝α_ｉＮ（μ_ｉ，σ_ｉ） （２）
式（２）において、ｉ＝１〜ｎであり、Ｎ（μ_ｉ，σ_ｉ）は正規分布の確率密度を示す。また、それぞれの局所周期成分Ｘ_ｔについての各局所パターンｉへの帰属確率Ｐ_ｔｉを以下の式（３）に基づいて算出する（サブステップＳＴ１２）。
【数２】

それぞれの局所周期成分Ｘ_ｔに係る帰属確率Ｐ_ｔｉが求まれば、以下の式（４）に基づいて、各局所パターンｉについて平均、分散、混合比から成る分類パラメータ（μ_ｉ，σ_ｉ，α_ｉ）を算出する（サブステップＳＴ１３）。なお、式（４）においては、対角共分散成分のみを計算する場合を例としている。
【数３】

また、式（４）により算出された平均、分散、混合比から成る分類パラメータ（μ_ｉ，σ_ｉ，α_ｉ）を記録するとともに、当該分類パラメータの変遷に係る履歴を参照して、所定の基準に基づいた履歴データに係る検証を実施し、当該検証結果に応じて次に実行するサブステップを選択する。例えば、それまでに実行された１０回の分類パラメータ算出工程にわたって平均、分散、混合比の値に係る変動幅がいずれも所定のしきい値を越えない場合には、データ帰属最尤推定サブステップＳＴ１４へ進み、それ以外の場合にはデータ帰属確率算出サブステップＳＴ１２へ復帰する。平均、分散、混合比から成る分類パラメータ（μ_ｉ，σ_ｉ，α_ｉ）に係る変化が所定のしきい値を越えることなく収束状態が確認されれば、最終的に得られた平均、分散、混合比から成る分類パラメータ（μ_ｉ，σ_ｉ，α_ｉ）を混合正規分布を規定する分類パラメータとして記録する。次に、当該分類パラメータを用いて、データ帰属確率算出サブステップＳＴ１２と同様の処理を実施して、それぞれの局所周期成分Ｘ_ｔについて各局所パターンｉへの帰属確率Ｐ_ｔｉを算出する。そして、それぞれの局所周期成分Ｘ_ｔ（ｔ＝１〜Ｔ）毎に、帰属確率Ｐ_ｔｉを比較することで、最大の帰属確率Ｐ_ｔｉを有する局所パターンに当該局所周期成分Ｘ_ｔが帰属するものと推定し、局所周期成分Ｘ_ｔと当該局所周期成分Ｘ_ｔが帰属する局所パターンとを対応付けて記録する（サブステップＳＴ１４）。
【００２３】
ステップＳＴ４においては、各局所パターンｉ毎に、当該局所パターンｉへの帰属が記録された局所周期成分Ｘ_ｔに属性データとして付加された識別子を調べて、当該局所パターンｉが特定の発信源種に固有な局所パターンとなり得るか否かについての判定を実施する。すなわち、局所パターンｉへの帰属が記録された局所周期成分Ｘ_ｔに属性データとして発信源識別子が付加されている場合には、発信源識別子の種類が単一であれば当該局所パターンｉが発信源識別子により明示される発信源種に固有の局所パターンであると判定する。また、発信源識別子が複数個混在しているのであれば、当該局所パターンｉへの帰属が記録された局所周期成分の数の最も多い発信源識別子に係る第１の発信源種と、記録された局所周期成分の数が２番目に多い発信源識別子に係る第２の発信源種とを比較して、第１の発信源種についての局所周期成分の記録数が第２の発信源種についての局所周期成分の記録数より所定のしきい値を越えて多い場合には、当該局所パターンｉが第１の発信源種に固有の局所パターンであると判定し、所定のしきい値を越えない場合には判定不能とする。以上のように、各局所パターンｉについて、当該局所パターンと当該局所パターンにより特定され得る発信源種とを対応付けることでそれぞれ個別の判定規則（以下、個別判定規則と称する）を得る。すべての局所パターンについて個別判定規則の存否が確認されれば、局所パターン毎に得られた個別判定規則および分類パラメータを組み合せて得られるデータを集成することでパルス識別判定規則を構成し、当該パルス識別判定規則をパルス判定部３へ出力する。
【００２４】
次に、図４は、パルス判定部における動作を示すフローチャートである。パルス判定部３は、レーダ波受信部１から未識別レーダ波データを入力すると、当該レーダ波データを適宜分断してパルスデータを取り出して格納する（ステップＳＴ２１）。次に、パルス判定部３は、取り出されたパルスデータを基にして、パルス波形に係る時系列データに対して所定の演算を適用することで変換を実施することにより、局所周期成分を時系列に抽出する（ステップＳＴ２２）。次に、パルス判定部３は、パルス識別判定規則を参照してステップＳＴ２２において生成されたそれぞれの局所周期成分の各局所パターンへの帰属状態に基づいて未識別レーダ波に係る発信源種を識別し、当該識別結果すなわちパルス識別結果を表示部５に出力する（ステップＳＴ２３）。
【００２５】
上記のステップＳＴ２１はパルス分類部２が実行するステップＳＴ１と同様であり、またステップＳＴ２２はパルス分類部２が実行するステップＳＴ２と同様であるので、その詳細な説明は省略する。ステップＳＴ２３においては、パルス分類部２から入力したパルス識別判定規則に含まれる分類パラメータに基づいて、未識別レーダ波データに係るそれぞれの局所周期成分Ｘ_ｔ毎に各局所パターンｉへの帰属確率を算出して、最大の帰属確率Ｐ_ｔｉを有する局所パターンｉに当該局所周期成分Ｘ_ｔが帰属するものと推定する。なお、帰属確率の算出は同様に式（３）を用いて実施する。次に、パルス識別判定規則を構成する各局所パターン毎の個別判定規則に基づいて、それぞれの局所周期成分が帰属する局所パターンのなかで発信源種との対応付けがなされている局所パターンを検出する。次に、発信源種との対応付けがなされている局所パターン毎に、当該局所パターンに帰属する局所周期成分の数を集計して、最も局所周期成分の帰属数の多い局所パターンに対応付けられた発信源種をもって受信した未識別レーダ波に係る発信源種であると判定し、当該判定結果をパルス識別結果として表示部５に出力する。また、受信されたレーダ波に係るそれぞれの局所周期成分が帰属する局所パターンのなかで発信源種との対応付けがなされている局所パターンがない場合、あるいは発信源種との対応付けがなされている局所パターンに帰属する局所周期成分の数が所定のしきい値より少ない場合には、判定不能とするパルス識別結果を表示部５に出力する。
【００２６】
次に、レーダ波諸元識別部４は、所与の諸元識別判定規則を参照して、レーダ波受信部１から入力する未識別レーダ波データに係る発信源種を識別し、当該識別結果すなわち諸元識別結果を表示部５へ出力する。なお、発信源種を識別するための基準となる電波諸元としては上述したように、周波数、ＰＲＦおよびパルス幅等の諸元値と、周波数モード、ＰＲＦモードおよびパルスモード等の諸元モードとがあり、これら電波諸元に基づくレーダ識別は例えば特開平１１−３２６４９４号公報に開示されているように周知技術であるので、ここではその説明を省略する。
【００２７】
次に、表示部５は、パルス判定部３からパルス識別結果を入力するとともに、レーダ波諸元識別部４から諸元識別結果を入力する。これら識別結果が得られれば、所与の総合識別判定規則を参照して、パルス識別結果および諸元識別結果から受信した未識別レーダ波に係る発信源種を識別して、当該識別結果をユーザに提示する。上記総合識別判定規則の一例について、以下に説明する。パルス識別結果と諸元識別結果とが一致した場合、あるいはパルス識別結果または諸元識別結果のいずれか一方が判定不能とされた場合には、パルス識別結果または諸元識別結果に基づいて特定される発信源種をユーザに提示する。また、パルス識別結果および諸元識別結果がともに判定不能とされた場合には、ユーザに対して判定不能と表示する。また、パルス識別結果と諸元識別結果とが一致しない場合には、いずれの識別結果を優先するかを予め規定しておいて、優先される識別結果に基づいて特定される発信源種をユーザに提示するか、あるいは、判定不能と表示するとともに双方の識別結果に基づいて特定される発信源種をともにユーザに提示する。なお、表示部５における発信源種等に係る表示形態を種々選択すること等を実施できるように、表示部５が表示形態を変更するための制御機能を有するのが好適である。
【００２８】
また、図５は、この発明の実施の形態１によるレーダ識別装置の変形例の構成を示すブロック図である。図５に示されるレーダ識別装置は、図１に示されたレーダ識別装置と比較すると、レーダ波諸元識別部４が削除されている点で相違する。すなわち、パルス判定部３から出力されるパルス識別結果のみを基にして、表示部５は受信されたレーダ波に係る発信源種あるいは識別不能である旨などをユーザに提示する。
【００２９】
以上のように、この実施の形態１によれば、レーダ波データを生成するレーダ波受信部１と、パルスデータに基づいて発信源種を識別するためのパルス識別判定規則を構成するパルス分類部２と、受信したレーダ波に係るパルスデータに係るそれぞれの局所周期成分についての各局所パターンへの帰属状態とパルス識別判定規則とを対照して発信源種を識別するパルス判定部３と、識別結果に基づいて受信されたレーダ波に係る発信源種などをユーザに提示する表示部５とを備えるように構成したので、レーダ波源となる装置を構成するマグネトロン等の構成部材についてそれぞれ固有の特徴が顕現するパルス波形に基づいて発信源種を識別するから、レーダ波の送受信において電波諸元に係る設定について過誤等が生じても正確な識別結果を得ることができ、識別精度を向上することができるという効果を奏する。
【００３０】
また、パルス分類部２が、レーダ波毎に時系列に得られるそれぞれの局所周期成分の各局所パターンへの帰属状態に基づいてパルス識別判定規則を構成するようにしたので、未識別レーダ波を含めた局所パターンに係る教師なし分類結果を利用することができて、学習用の識別付レーダ波が少なくても有効なパルス識別判定規則が得られて所望の識別精度を確保することが可能となり、実用性を向上することができるという効果を奏する。
【００３１】
また、パルス分類部２が、それぞれ正規分布として与えられる複数の局所パターン全体を表現する混合正規分布に基づいて、それぞれの局所周期成分の各局所パターンへの帰属確率を求めることで、当該局所周期成分の帰属する局所パターンを決定するように構成したので、局所パターンの分類を統計的に有効な手法を用いて実施できるから、パルス識別判定規則の信頼性の向上を図ることが可能となり、発信源種の識別精度を向上することができるという効果を奏する。
【００３２】
また、受信したレーダ波から電波諸元を抽出して、抽出された電波諸元に基づいて発信源種を識別するレーダ波諸元識別部４を備え、表示部５がパルス判定部３から出力されるパルス判定結果およびレーダ波諸元識別部４から出力される諸元識別結果に基づいて、受信されたレーダ波に係る発信源種をユーザに提示するように構成したので、パルス波形のみではなく種々の電波諸元にも基づいて発信源種を識別することができるから、識別精度をより向上することができるという効果を奏する。
【００３３】
実施の形態２．
この実施の形態２によるレーダ識別装置は、装置構成については実施の形態１によるレーダ識別装置と同様の構成を有しており、パルス分類部におけるパルス識別判定規則の構成に係る動作、およびパルス判定部における未識別レーダ波に係る識別動作が異なる点で実施の形態１と相違する。
【００３４】
以下、この実施の形態２によるレーダ識別装置に係る動作について説明する。なお、レーダ波受信部１、レーダ波諸元識別部４および表示部５に係る動作については実施の形態１と同様であるので、その説明を省略する。図６は、この発明の実施の形態２によるレーダ識別装置に係るパルス分類部における動作を示すフローチャートである。パルス分類部２は、レーダ波受信部１からレーダ波データを入力すると、実施の形態１においてパルスデータに対して実施したのと同様に、レーダ波データに対して所定の演算を適用することで局所的な特徴量への変換を実施して局所周期成分を時系列に抽出する（ステップＳＴ３１）。次に、パルス分類部２は、レーダ波について時系列に与えられる局所周期成分がそれぞれ帰属し得る複数の局所パターンおよび局所パターン間の遷移（自己回帰を含む）を規定する複数の遷移パターンから成り、レーダ波に係る特徴量の時間的変化を表現するモデル（以下、状態遷移モデルと称する）に基づいて、レーダ波毎に時系列に得られる局所周期成分が状態遷移モデル上においてその特徴量変化に応じて辿る経路（以下、遷移経路と称する）のなかで最も尤度の高い遷移経路を推定するとともに、状態遷移モデルにおいてレーダ波全体に係る変化状態を規定するモデルパラメータ（モデルパラメータについては後述する）を算出する（ステップＳＴ３２）。なお、任意のレーダ波について時系列に与えられた局所周期成分の変化に基づいて与えられる遷移経路において、時間的に隣接する局所周期成分の変化に対して設定されて上記遷移経路の一部を構成する経路を部分経路と称するものとする。また、それぞれの部分経路は、当該部分経路の設定をなす時間的に隣接する局所周期成分がそれぞれ帰属する局所パターン間に規定された遷移パターンに帰属するものとする。また、部分経路は、上述したように、それぞれ時系列に設定されるものであるから、複数の部分経路が同一の遷移パターンに帰属することが当然に予想され、このような部分経路の同一遷移パターンへの重複度もレーダ波に係るそれぞれの部分経路についての各遷移パターンへの帰属状態を表現する重要な要素となる。次に、パルス分類部２は、識別付レーダ波毎に時系列に得られる局所周期成分について隣接する局所周期成分間に設定されるそれぞれの部分経路の各遷移パターンへの帰属状態に基づいてパルス識別判定規則を生成する（ステップＳＴ３３）。なお、パルス分類部２によるパルス識別判定規則の構成を監視並びに制御するとともに、パルス識別判定規則を修正できるように、パルス分類部２が表示機能および制御機能を有するのが好適である。
【００３５】
次に、上記のステップＳＴ３１からステップＳＴ３３として示された各工程について詳細に説明する。ステップＳＴ３１においては、実施の形態１においてパルスデータに対して実施したのと同様に、レーダ波受信部１から入力するレーダ波データに対して所定の演算を適用することで局所的な特徴量への変換を実施して、局所周期成分を時系列に算出するとともに、これら局所周期成分を所定の記憶手段に格納する。
【００３６】
ステップＳＴ３２においては、上述したように所与の状態遷移モデルに基づいて、レーダ波毎に与えられる局所周期成分の時系列データに基づいてモデルパラメータを算出する。図７は、状態遷移モデルの一例を示す図である。図７において、Ａ１はパルス間状態、Ａ２はパルス開始状態、Ａ３はパルス状態、Ａ４はパルス終了状態、Ｂｉ（ｉ＝１〜１３）はそれぞれの局所周期成分が何れかに帰属し得る局所パターン、Ｒ_ｉ（ｉ＝１〜１１）はレーダ波データの特性に基づいて局所パターン間での可能な遷移を規定する遷移パターンである。図７に示されるように、この状態遷移モデルは、パルス間状態Ａ１、パルス開始状態Ａ２、パルス状態Ａ３およびパルス終了状態Ａ４の４状態から成るループ状の状態遷移を想定するモデルとして与えられている。各状態は、自己回帰となる遷移パターンをも有するとともに、それぞれ任意数の局所パターンを有している。例えば、パルス開始状態Ａ２の有する局所パターンＢ３，Ｂ４，Ｂ５は、相互に全結合の遷移ネットワークを形成するとともに、それぞれがパルス間状態Ａ１からの遷移パターンＲ２およびパルス状態Ａ３への遷移パターンＲ７を有している。なお、遷移パターンＲ２およびＲ７に関しては、局所パターンＢ３，Ｂ４，Ｂ５についてまとめて１つの遷移パターンを設定するのではなく、局所パターンＢ３，Ｂ４，Ｂ５毎にそれぞれ３つの遷移パターンを設定するようにしてもよい。また、局所周期成分の帰属するそれぞれの局所パターンは例えば単一の正規分布で表現することが可能である。これら局所パターン間の遷移は確率的に定まり、当該確率を遷移確率と称することとする。さらに、各局所パターンについて、当該局所パターンを表現する正規分布パラメータと、当該局所パターンから延びる各遷移パターンに係る遷移確率とを組み合せて構成されるパラメータをモデルパラメータと称することとする。
【００３７】
図８は、状態遷移モデルにおけるモデルパラメータ算出工程を示すフローチャートである。ここで、レーダ波データ毎に時系列に得られるＴ個の局所周期成分をＸ_１〜Ｘ_Ｔとするとともに、局所パターン数をｎ個とする。まず、モデルパラメータを以下に示す期待値最大化法で求めるための初期値を設定する（サブステップＳＴ４１）。ｎ個の局所パターンについて、例えばレーダ波毎に時系列に与えられる局所周期成分Ｘ_１〜Ｘ_Ｔからｎ個の互いに異なるものを適宜選択してそれぞれ各局所パターンについての平均の初期値とし、Ｘ_１〜Ｘ_Ｔから分散を算出してｎ個の局所パターンに共通な分散の初期値とし、遷移確率を等確率とする。すなわち、遷移確率については、任意の局所パターンから４つの遷移パターンが延びているとすれば、それぞれの遷移パターンに係る遷移確率を０．２５とする。各局所パターンについての平均、分散、遷移確率の組から成るモデルパラメータ（μ_ｉ，σ_ｉ，α_ｉ）が与えられれば、各レーダ波毎に当該レーダ波に係る局所周期成分Ｘ_１〜Ｘ_Ｔについて最も尤度の高い遷移経路を検出する（サブステップＳＴ４２）。各レーダ波に係る遷移経路が検出されれば、当該検出結果に基づいて各局所パターンについて平均、分散、遷移確率の組から成るモデルパラメータ（μ_ｉ，σ_ｉ，α_ｉ）を算出する（サブステップＳＴ４３）。サブステップＳＴ４３において、モデルパラメータが算出されると、当該モデルパラメータの変遷に係る履歴を参照して、所定の基準に基づいた履歴データに係る検証を実施して当該検証結果に応じて次に実行するサブステップを選定する。例えば、それまでに実行された１０回のモデルパラメータ算出工程にわたっての平均、分散、遷移確率の値に係る変動幅が所定のしきい値を越える場合には、遷移経路最尤推定サブステップＳＴ４２へ復帰し、それ以外の場合には、平均、分散、遷移確率から成るモデルパラメータ（μ_ｉ，σ_ｉ，α_ｉ）が収束状態にあるとみなして、最終的に得られたモデルパラメータ（μ_ｉ，σ_ｉ，α_ｉ）を状態遷移モデルを確定するモデルパラメータとして記録する。
【００３８】
ステップＳＴ３３においては、レーダ波毎に、パルス間状態Ａ１に含まれる局所パターン間での遷移を省くことで遷移経路を分断することにより、パルス波形に対応する遷移経路（以下、パルス遷移経路と称する）を抽出する。次に、それぞれの識別付レーダ波について、当該識別付レーダ波に係るパルス遷移経路を辿って通過した部分経路が帰属する遷移パターンに当該発信源識別子を属性データとして付属させる。すべての識別付レーダ波について遷移パターンへの発信源識別子の付属処理が終了すれば、各遷移パターン毎に、当該遷移パターンへの付属が記録された発信源識別子を調べて、当該遷移パターンが特定の発信源種に固有な遷移パターンとなり得るか否かについて判定を実施する。発信源識別子の付属が記録された任意の遷移パターンについて、記録された発信源識別子の種類が単一であれば当該遷移パターンが発信源識別子により明示される発信源種に固有の遷移パターンであると判定する。また、発信源識別子が複数個混在しているのであれば、当該遷移パターンへ付属させられた記録数の最も多い発信源識別子に係る第１の発信源種と、記録数の２番目に多い発信源識別子に係る第２の発信源種とを比較して、第１の発信源種についての記録数が第２の発信源種についての記録数より所定のしきい値を越えて多い場合には当該遷移パターンが第１の発信源種に固有の遷移パターンであると判定し、所定のしきい値を越えない場合には判定不能とする。以上のように、各遷移パターンについて、当該遷移パターンと当該遷移パターンにより特定され得る発信源種とを対応付けることでそれぞれ個別の判定規則（以下、個別判定規則と称する）を得る。すべての遷移パターンについて個別判定規則の存否が確認されれば、遷移パターン毎に得られた個別判定規則およびモデルパラメータを組み合せて得られるデータを集成することでパルス識別判定規則を構成し、当該パルス識別判定規則をパルス判定部３へ出力する。
【００３９】
次に、パルス判定部３の動作について説明する。パルス判定部３は、レーダ波受信部１から未識別レーダ波に係るレーダ波データを入力すると、当該レーダ波データを基にして、レーダ波形に係る時系列データに対してパルス分類部２で用いたのと同様の所定の演算を適用することで変換を実施して、局所周期成分の時系列データを算出する。次に、パルス判定部３は、各局所パターン毎に確定された平均、分散、遷移確率の組から成るモデルパラメータ（μ_ｉ，σ_ｉ，α_ｉ）に基づいて、当該レーダ波データに係る局所周期成分の時系列データＸ_１〜Ｘ_Ｔについて最も尤度の高い遷移経路を検出する。最も尤度の高い遷移経路が推定されれば、当該遷移経路を辿って通過する部分経路が帰属する遷移パターンについて発信源種との対応付けがなされている場合には、当該発信源識別子を登録する。登録された発信源識別子が１つの場合には、当該発信源識別子により明示される発信源種をもって受信した未識別レーダ波に係る発信源種であると判定して当該パルス識別結果を表示部５に出力する。また、登録された発信源識別子が複数の場合には、対応付けられる遷移パターン数が最も多い発信源識別子により明示される発信源種をもって受信した未識別レーダ波に係る発信源種であると判定して当該パルス識別結果を表示部５に出力する。また、登録された発信源識別子がない場合、あるいは登録された発信源識別子が複数あっていずれの発信源識別子についても対応付けられる遷移パターン数が所定のしきい値より少ない場合には、判定不能とするパルス識別結果を表示部５に出力する。
【００４０】
以上のように、この実施の形態２によれば、レーダ波データを生成するレーダ波受信部１と、レーダ波データに係る局所周期成分についての状態遷移モデル上での遷移経路に基づいて発信源種を識別するためのパルス識別判定規則を構成するパルス分類部２と、受信したレーダ波について推定される遷移経路とパルス識別判定規則とを対照して発信源種を識別するパルス判定部３と、識別結果に基づいて受信されたレーダ波に係る発信源種をユーザに提示する表示部５とを備えるように構成したので、レーダ波源となる装置を構成するマグネトロン等の構成部材についてそれぞれ固有の特徴が顕現するパルス波形に基づいて発信源種を識別することができるから、レーダ波の送受信において電波諸元に係る設定について過誤等が生じても正確な識別結果を得ることができ、識別精度を向上することができるという効果を奏する。
【００４１】
また、レーダ波諸元識別部４を備えることで、実施の形態１に記載したと同じ効果を奏する。さらに、局所周期成分についての状態遷移モデル上での遷移経路に基づいて発信源種を識別するので、パルス波形に対応する局所周期成分の集合を容易に同定することができ、レーダ波データからパルス波形に係る特徴を抽出するための基準値設定等の予備作業を省くことが可能となり、ユーザの負担を軽減して使い勝手を向上することができるという効果を奏する。
【００４２】
実施の形態３．
この実施の形態３によるレーダ識別装置は、装置構成については実施の形態２によるレーダ識別装置と同様の構成を有しており、パルス分類部におけるパルス識別判定規則の構成に係る動作が異なる点で実施の形態２と相違する。
【００４３】
以下、この実施の形態３によるレーダ識別装置に係る動作について説明する。なお、レーダ波受信部１、パルス判定部３、レーダ波諸元識別部４および表示部５に係る動作については実施の形態２と同様であるので、その説明を省略する。図９は、この発明の実施の形態３によるレーダ識別装置に係るパルス分類部における動作を示すフローチャートである。パルス分類部２は、レーダ波受信部１からレーダ波データを入力すると、実施の形態２と同様にレーダ波に係る時系列データに対して所定の演算を適用することで局所的特徴量への変換を実施して局所周期成分の時系列データを算出する（ステップＳＴ５１）。次に、パルス分類部２は、状態遷移モデルに基づいて、レーダ波毎に時系列に得られる局所周期成分について最も尤度の高い遷移経路を推定するとともに、状態遷移モデルにおいてレーダ波全体に係る変化状態を規定するモデルパラメータを算出する（ステップＳＴ５２）。次に、パルス分類部２は、識別付レーダ波毎に時系列に得られる局所周期成分について隣接する局所周期成分間に設定されるそれぞれの部分経路の各遷移パターンへの帰属状態に基づいて、所定の条件が満たされていればパルス識別判定規則を生成する（ステップＳＴ５３）。また、所定の条件が満たされない場合には、レーダ波に係る特徴量の時間的変化についてモデル化の精度を高めるために状態遷移モデルを変形する（ステップＳＴ５４）。なお、パルス分類部２によるパルス識別判定規則の構成を監視並びに制御するとともに、パルス識別判定規則を修正できるように、パルス分類部２が表示機能および制御機能を有するのが好適である。
【００４４】
次に、上記のステップＳＴ５１からステップＳＴ５４として示された各工程について詳細に説明する。なお、ステップＳＴ５１については、実施の形態２におけるステップＳＴ３１と同様であるのでその説明を省略する。また、ステップＳＴ５２については、設定される状態遷移モデルの形態が変化する点でのみ実施の形態２におけるステップＳＴ３２と相違し、基本的な動作についてはステップＳＴ３２と同様であるのでその説明を省略する。すなわち、予め初期モデルとして設定された状態遷移モデルに基づいてモデルパラメータを算出する場合もあれば、状態遷移モデル変形ステップＳＴ５４により変形された状態遷移モデルに基づいてモデルパラメータを算出する場合もある。
【００４５】
ステップＳＴ５３においては、レーダ波毎に、パルス間状態Ａ１に含まれる局所パターン間での遷移を省くことで遷移経路を分断することにより、パルス波形に対応するパルス遷移経路を抽出する。次に、それぞれの識別付レーダ波について、当該識別付レーダ波に係るパルス遷移経路を辿って通過したそれぞれの部分経路が帰属する遷移パターンに当該識別付レーダ波に係る発信源識別子を属性データとして付属させる。すべての識別付レーダ波について遷移パターンへの発信源識別子の付属処理が終了すれば、各遷移パターン毎に、当該遷移パターンへの付属が記録された発信源識別子を調べて、当該遷移パターンが特定の発信源種に固有な遷移パターンとなり得るか否かについて判定を実施して個別判定規則を生成する。なお、特定の発信源種に固有な遷移パターンとなり得るか否かに係る判定処理については、実施の形態２のステップＳＴ３３で説明済みであるので、ここではその説明を省略する。そして、すべての遷移パターンについて個別判定規則の存否が確認されれば、所定数のパルス遷移経路について、パルス遷移経路毎に、当該パルス遷移経路を辿る際に通過したそれぞれの部分経路が帰属する遷移パターンへの付属が記録された発信源識別子の種類を計数する。発信源識別子の種類数すなわち発信源種の種類数が０または２以上である場合、あるいは当該判定規則決定ステップＳＴ５３の実行回数が規定回数以下である場合には、状態遷移モデル変形ステップＳＴ５４を実行する。また、発信源種の種類数が１である場合、あるいは当該判定規則決定ステップＳＴ５３の実行回数が規定回数を越えた場合には、遷移パターン毎に個別判定規則およびモデルパラメータを組み合わせて得られるデータを集成することでパルス識別判定規則を構成し、当該パルス識別判定規則をパルス判定部３へ出力する。
【００４６】
ステップＳＴ５４においては、局所パターン毎にモデルパラメータを検索して最も分散値の大きな局所パターンを検出し、当該局所パターンを分割して２つの新たな局所パターンを生成する。この２つの局所パターン間における遷移は全結合とする。また、それぞれの局所パターンに係るモデルパラメータの初期値については、例えば一方の局所パターンには分割前の局所パターンのモデルパラメータを付与し、他方の局所パターンには所与の微小量を上限として発生された乱数を基にして分割前の局所パターンのモデルパラメータを修正して得たモデルパラメータを付与する。
【００４７】
以上のように、この実施の形態３によれば、実施の形態２と同等の効果を奏するとともに、パルス分類部２が、任意のレーダ波に係る遷移経路を検証して、当該遷移経路を構成する個々の部分経路が帰属するそれぞれの遷移パターンにより特定し得る発信源種の種類数が０または２以上である場合に状態遷移モデルを変形するように構成したので、自動的に最適な状態遷移モデルを構成することができるから、予め精度の良いモデル化を実施する必要がなく、ユーザの負担を軽減することができるという効果を奏する。
【００４８】
実施の形態４．
図１０は、この発明の実施の形態４によるレーダ識別装置の構成を示すブロック図である。この実施の形態４によるレーダ識別装置は、識別実施形態で使用される機能要素から携帯器を分離独立して構成する点で実施の形態１から実施の形態３と相違する。図１０において、図１と同一符号は同一または相当部分を示すのでその説明を省略する。１１はパルス分類部２を有して構成される基地局、１２はレーダ波受信部１、パルス判定部３、レーダ波諸元識別部４および表示部５を有して構成される携帯器である。
【００４９】
次に動作について説明する。なお、基本的な動作は実施の形態１または実施の形態３のいずれかによるレーダ識別装置と同様であるので、ここでは実施の形態４によるレーダ識別装置について特徴的な動作についてのみ説明するものとする。識別準備形態においては、１または複数の携帯器１２内のレーダ受信部１は、基地局１１と携帯器１２との間の通信回線が確立されている場合には受信したレーダ波に係るレーダ波データを基地局１１内のパルス分類部２へ送信し、基地局１１と携帯器１２との間の通信回線が確立されていない場合にはレーダ波データを記録するとともに通信回線が確立された時点で記録されていたレーダ波データを基地局１１内のパルス分類部２へ送信する。パルス分類部２は、蓄積された複数のパルスデータに基づいてパルス識別判定規則を構成して当該パルス識別判定規則を基地局１１と携帯器１２との間の通信回線が確立されている間に１または複数の携帯器１２内のパルス判定部３へ送信する。また、識別実施形態においては、レーダ波受信部１とパルス分類部２とに係る動作は上記と同様であるとともに、それぞれの携帯器１２は、当該携帯器１２と基地局１１との間の通信回線の確立の有無に関わらず、それぞれ個別にレーダ波を受信して当該レーダ波に係る発信源種を識別する。
【００５０】
以上のように、この実施の形態４によれば、実施の形態１から実施の形態３のいずれかと同等の効果を奏するとともに、パルス分類部２を有して構成される基地局１１と、レーダ波受信部１、パルス判定部３、レーダ波諸元識別部４および表示部５を有して構成される携帯器１２とを備えるように構成したので、携帯器１２のみで識別実施形態での使用を独立に実現することができるから、未識別レーダ波に係る発信源種の識別を実施する識別装置を小型軽量化することができるという効果を奏する。また、複数の携帯器１２について基地局１１を共有することができるから、レーダ識別を実施するシステム全体のコスト低減を図れるという効果を奏する。
【００５１】
なお、レーダ識別装置が実施の形態１の変形例に示すようにレーダ波諸元識別部４を具備しない構成を有する場合には、パルス分類部２を有して構成される基地局１１と、レーダ波受信部１、パルス判定部３および表示部５を有して構成される携帯器１２とを備える構成とすることで、上記と同様の効果を奏することができる。
【００５２】
実施の形態５．
図１１は、この発明の実施の形態５によるレーダ識別装置の構成を示すブロック図である。この実施の形態５によるレーダ識別装置は、自律稼動可能な自動器にレーダ波受信部を備える点で実施の形態４と相違する。図１１において、図１と同一符号は同一または相当部分を示すのでその説明を省略する。２１はレーダ波受信部１を有して構成され好ましくは自律稼動可能である自動器、２２はパルス分類部２を有して構成される基地局、２３はパルス判定部３、レーダ波諸元識別部４および表示部５を有して構成される携帯器である。なお、自動器２１としては、例えば小型のラジコンヘリのような無線操作または自動操縦で飛行する航空機などが想定される。
【００５３】
次に動作について説明する。なお、基本的な動作は実施の形態１から実施の形態３のいずれかによるレーダ識別装置と同様であるので，ここでは実施の形態５によるレーダ識別装置について特徴的な動作についてのみ説明するものとする。識別準備形態においては、自動器２１内のレーダ波受信部１は、自動器２１と基地局２２との間の通信回線が確立されている場合には受信したレーダ波に係るレーダ波データを基地局２２内のパルス分類部２へ送信し、自動器２１と基地局２２との間の通信回線が確立されていない場合にはレーダ波データを記録するとともに通信回線が確立された時点で記録されていたレーダ波データを基地局２２内のパルス分類部２へ送信する。パルス分類部２は、蓄積された複数のパルスデータに基づいてパルス識別判定規則を構成して当該パルス識別判定規則を基地局２２と携帯器２３との間の通信回線が確立されている間に１または複数の携帯器２３内のパルス判定部３へ送信する。また、識別実施形態においては、レーダ波受信部１とパルス分類部２とに係る動作は上記と同様であるとともに、それぞれの携帯器２３は、自動器２１と携帯器２３との間の通信回線が確立されている間において、それぞれ個別にレーダ波を受信して当該レーダ波に係る発信源種を識別する。
【００５４】
以上のように、この実施の形態５によれば、実施の形態１から実施の形態３のいずれかと同等の効果を奏するとともに、レーダ波受信部１を有して構成され好ましくは自律稼動可能である自動器２１と、パルス分類部２を有して構成される基地局２２と、パルス判定部３、レーダ波諸元識別部４および表示部５を有して構成される携帯器２３とを備えるように構成したので、自動器２１から受信されたレーダ波に係るレーダ波データが送信されれば、携帯器２３のみで識別実施形態での使用を独立に実現することができるから、未識別レーダ波に係る発信源種の識別を実施する識別装置を小型軽量化できるという効果を奏する。また、複数の携帯器２３について自動器２１および基地局２２を共有することができるから、レーダ識別を実施するシステム全体のコスト低減を図れるという効果を奏する。
【００５５】
なお、レーダ識別装置が実施の形態１の変形例に示すようにレーダ波諸元識別部４を具備しない構成を有する場合には、レーダ波受信部１を有して構成される自動器２１と、パルス分類部２を有して構成される基地局２２と、パルス判定部３および表示部５を有して構成される携帯器２３とを備える構成とすることで、上記と同様の効果を奏することができる。
【００５６】
また、上記の実施の形態１から実施の形態５により開示されたレーダ識別装置は本願発明を例示するものであり、その技術的範囲を限定するものではないことに留意されたい。本願発明の技術的範囲は特許請求の範囲の記載に基づいて定められ、特許請求の範囲と均等な技術的思想および特許請求の範囲内における種々の設計的変更が本願発明に含まれるものである。
【００５７】
【発明の効果】
以上のように、この発明によれば、レーダ波を受信してレーダ波データを生成するレーダ波受信部と、パルスデータを生成するとともに当該パルスデータに係る局所周期成分を抽出し、レーダ波毎のパルスデータについて得られるそれぞれの局所周期成分を所定数に分類された複数の局所パターンのなかのいずれかの局所パターンに帰属させ、発信源種が予め明らかになっているレーダ波に係る局所周期成分の各局所パターンへの帰属状態に基づいてパルス識別判定規則を構成するパルス分類部と、受信されたレーダ波に係る局所周期成分の各局所パターンへの帰属状態とパルス判定識別規則とを対照することにより発信源種を識別するパルス判定部と、受信されたレーダ波に係る発信源種をユーザに提示する表示部とを備えるように構成したので、レーダ波源となる装置を構成するマグネトロン等の構成部材についてそれぞれ固有の特徴が顕現するパルス波形に基づいて発信源種を識別するから、レーダ波の送受信において電波諸元に係る設定について過誤等が生じても正確な識別結果を得ることができ、識別精度を向上することができるという効果を奏する。また、パルス分類部がレーダ波毎に時系列に得られるそれぞれの局所周期成分の各局所パターンへの帰属状態に基づいてパルス識別判定規則を構成するようにしたので、未識別レーダ波をも含めた局所パターンに係る教師なし分類結果を利用することができるから、学習用の識別付レーダ波が少なくても有効なパルス識別判定規則が得られて所望の識別精度を確保することが可能となり、実用性を向上することができるという効果を奏する。
【００５８】
この発明によれば、パルス分類部が、それぞれ正規分布として与えられる複数の局所パターン全体を表現する混合正規分布に基づいて、それぞれの局所周期成分の各局所パターンへの帰属確率を求めることで、当該局所周期成分の帰属する局所パターンを決定するように構成したので、局所パターンの分類を統計的に有効な手法を用いて実施でき、パルス識別判定規則の信頼性の向上を図ることが可能となるから、発信源種の識別精度を向上することができるという効果を奏する。
【００５９】
この発明によれば、レーダ波を受信してレーダ波データを生成するレーダ波受信部と、当該レーダ波データに係る局所周期成分を抽出し、状態遷移モデルに基づいてレーダ波毎に当該レーダ波に係る局所周期成分が状態遷移モデル上で遷移する遷移経路を推定し、発信源種が予め明らかになっているレーダ波の遷移経路を構成する個々の部分経路についての各遷移パターンへの帰属状態に基づいてパルス識別判定規則を構成するパルス分類部と、受信されたレーダ波について推定された遷移経路を構成する個々の部分経路の各遷移パターンへの帰属状態とパルス識別判定規則とを対照することにより発信源種を識別するパルス判定部と、受信されたレーダ波に係る発信源種をユーザに提示する表示部とを備えるように構成したので、レーダ波源となる装置を構成するマグネトロン等の構成部材についてそれぞれ固有の特徴が顕現するパルス波形に基づいて発信源種を識別するから、レーダ波の送受信において電波諸元に係る設定について過誤等が生じても正確な識別結果を得ることができ、識別精度を向上することができるという効果を奏する。また、レーダ波の局所周期成分に係る遷移経路に基づいて発信源種を識別するので、パルス波形に対応する局所周期成分の集合を容易に同定することができ、レーダ波データからパルス波形に係る特徴を抽出するための基準値設定等の予備作業を省くことが可能となり、ユーザの負担を軽減して使い勝手を向上することができるという効果を奏する。
【００６０】
この発明によれば、パルス分類部が、任意のレーダ波に係る局所周期成分の遷移経路を検証して、当該遷移経路を構成する個々の部分経路が帰属するそれぞれの遷移パターンにより特定し得る発信源種の種類数が０または２以上である場合に、状態遷移モデルを変形するように構成したので、自動的に最適な状態遷移モデルを構成することが可能となり、予め精度の良いモデル化を実施する必要がなく、ユーザの負担を軽減して使い勝手を向上することができるという効果を奏する。
【００６１】
この発明によれば、レーダ波データから１または複数の電波諸元を抽出し、抽出された電波諸元に基づいて受信されたレーダ波に係る発信源種を識別するレーダ波諸元識別部を備え、表示部が、パルス判定部から出力される識別結果とレーダ波諸元識別部から出力される識別結果とに基づいて、受信されたレーダ波に係る発信源種をユーザに提示するように構成したので、パルス波形のみではなく種々の電波諸元にも基づいて発信源種を識別することができるから、識別精度をより向上することができるという効果を奏する。
【００６２】
この発明によれば、パルス分類部を有して構成される基地局と、少なくともレーダ波受信部、パルス判定部および表示部を有して構成される携帯器とを備えるように構成したので、携帯器のみで識別実施形態での使用を独立に実現することができるから、未識別レーダ波に係る発信源種を識別する装置を小型軽量化することができるという効果を奏する。また、複数の携帯器について基地局を共有することができるから、レーダ識別を実施するシステム全体のコスト低減を図れるという効果を奏する。
【００６３】
この発明によれば、レーダ波受信部を有して構成される自動器と、パルス分類部を有して構成される基地局と、少なくともパルス判定部および表示部を有して構成される携帯器とを備えるように構成したので、自動器から受信されたレーダ波に係るレーダ波データが送信されれば、携帯器のみで識別実施形態での使用を独立に実現することができるから、未識別レーダ波に係る発信源種を識別する装置を小型軽量化することができるという効果を奏する。また、複数の携帯器について自動器および基地局を共有することができるから、レーダ識別を実施するシステム全体のコスト低減を図れるという効果を奏する。
【図面の簡単な説明】
【図１】 この発明の実施の形態１によるレーダ識別装置の構成を示すブロック図である。
【図２】 この発明の実施の形態１によるレーダ識別装置におけるパルス分類部に係る動作を示すフローチャートである。
【図３】 教師なし分類動作を示すフローチャートである。
【図４】 この発明の実施の形態１によるレーダ識別装置におけるパルス判定部に係る動作を示すフローチャートである。
【図５】 この発明の実施の形態１によるレーダ識別装置の変形例の構成を示すフローチャートである。
【図６】 この発明の実施の形態２によるレーダ識別装置におけるパルス分類部に係る動作を示すフローチャートである。
【図７】 この発明の実施の形態２によるレーダ識別装置において使用される状態遷移モデルの一例を示す図である。
【図８】 モデルパラメータの算出工程を示すフローチャートである。
【図９】 この発明の実施の形態３によるレーダ識別装置におけるパルス分類部に係る動作を示すフローチャートである。
【図１０】 この発明の実施の形態４によるレーダ識別装置の構成を示すブロック図である。
【図１１】 この発明の実施の形態５によるレーダ識別装置の構成を示すブロック図である。
【図１２】 従来のレーダ識別装置の構成を示すブロック図である。
【符号の説明】
１ レーダ波受信部、２ パルス分類部、３ パルス判定部、４ レーダ波諸元識別部、５ 表示部、１１，２２ 基地局、１２，２３ 携帯器、２１ 自動器、Ａ１ パルス間状態、Ａ２ パルス開始状態、Ａ３ パルス状態、Ａ４ パルス終了状態、Ｂ１〜Ｂ１３ 局所パターン、Ｒ１〜Ｒ１１ 遷移パターン。[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a radar identification device that identifies a source type of a radar wave based on radar wave data obtained for a received radar wave, and more particularly to a radar identification device that also employs a pulse shape characteristic as an identification reference. Is.
[0002]
[Prior art]
FIG. 12 is a block diagram showing a configuration of a conventional radar identification device. In FIG. 12, reference numeral 101 denotes a radar wave receiving unit that receives radar waves and generates radar wave data, and reference numeral 102 extracts various radio wave specifications from the radar wave data input from the radar wave receiving unit 101, and extracted. By comparing the radio wave specifications with a given specification identification dictionary, the type and type of the transmission source related to the received radar wave (which identifies the radar site that transmits the radar wave, etc. Is a radar wave item identification unit that identifies a source type and outputs an identification result, and 103 is a display unit that presents the identification result input from the radar wave item identification unit 102 to the user.
[0003]
As the radio wave parameters extracted by the radar wave parameter identifying unit 102, for example, according to the radio wave source name identifying device disclosed in JP-A-11-326494, frequency, PRF (Pulse Repetition Frequency), pulse width, etc. The frequency mode for discriminating whether the frequency is always constant or fluctuating within a predetermined range, and the pulse output interval for PRF is always constant or fluctuating within a predetermined range And a specification mode such as a pulse mode for discriminating whether a pulse output at each repetition interval is 1 or 2 is used. Further, in the radio wave source name identification device disclosed in the above publication, the radar wave received for each radio wave specification indicates the possibility that it corresponds to each transmission source type by referring to a given specification identification dictionary. A probability score is calculated, and the source type related to the received radar wave is identified based on the total of the probability scores. In the radar identification device disclosed in Japanese Patent Laid-Open No. 9-264946, a specification group effective for identification is established in a specification identification dictionary in order to identify the source type at high speed.
[0004]
[Problems to be solved by the invention]
Since the conventional radar identification apparatus is configured as described above, as described above, the source type related to the received radar wave is identified based on radio wave specifications such as frequency, PRF, and pulse width. However, these radio wave specifications are set based on a predetermined rule for each transmission source type, so the radio wave specifications and reception set on the transmission side in the transmission and reception of radar waves due to some error etc. There is a possibility that the radio wave specification set on the side may not match in whole or in part, and an accurate identification result cannot always be obtained. In addition, when sufficient specification identification dictionary covering many source types is not provided in advance, or when learning data for identifying source types is not sufficiently obtained, the discrimination performance is improved. There was a problem of being low.
[0005]
The present invention has been made to solve the above-described problems, and can obtain a radar identification device with improved identification accuracy by using an analysis related to a pulse waveform in which a characteristic unique to a radar wave source appears. With the goal.
[0006]
Moreover, even if sufficient identification dictionaries related to radar waves cannot be obtained in advance or sufficient data for learning cannot be obtained, practical use is possible by unsupervised learning including radar waves whose source type is not clear. An object of the present invention is to obtain a radar identification device that can ensure a certain identification accuracy.
[0007]
[Means for Solving the Problems]
A radar identification device according to the present invention includes a radar wave receiving unit that receives radar waves and generates radar wave data, generates pulse data by appropriately dividing the radar wave data, and generates local feature amounts from the pulse data. Extracting the local periodic component given in time series as and assigning each local periodic component obtained for the pulse data for each radar wave to any one of a plurality of local patterns classified into a predetermined number, The transmission source type and the local pattern that can identify the transmission source type are associated with each other based on the attribution state of the local periodic component related to the radar wave whose transmission source type is known in advance to each local pattern. Pulses constituting a pulse identification judgment rule for identifying source types based on pulse data by assembling combinations of different source types and local patterns The pulse data is generated by appropriately dividing the class part and the radar wave data, the local periodic component is extracted from the pulse data, and each local periodic component is included in the same plurality of local patterns set by the pulse classifying unit. Identify the source type and output the identification result by making it belong to one of the local patterns, comparing the attribution state of each local periodic component related to the received radar wave to each local pattern and the pulse identification judgment rule A pulse determination unit and a display unit for presenting to the user that the type of transmission source related to the received radar wave or indistinguishment based on the identification result to be input are provided.
[0008]
In the radar identification device according to the present invention, the pulse classification unit obtains the belonging probability of each local periodic component to each local pattern based on a mixed normal distribution that represents the entire plurality of local patterns each given as a normal distribution. Thus, the local pattern to which the local periodic component belongs is determined.
[0009]
A radar identification device according to the present invention includes a radar wave receiving unit that receives a radar wave and generates radar wave data, and extracts a local periodic component given in time series as a local feature amount from the radar wave data. For each radar wave based on a state transition model consisting of a plurality of local patterns to which a local periodic component given in time series can belong and a plurality of transition patterns that define transitions between local patterns. Based on the attribution state to each transition pattern for each partial path that constitutes the transition path of the radar wave whose period component transitions on the state transition model and whose source type is known in advance Associating a source type with a transition pattern that can identify the source type, and assembling a combination of the associated source type and transition pattern A pulse classification unit that constitutes a pulse identification judgment rule for identifying a source type based on radar wave data, and a radar wave received based on a state transition model by extracting a local periodic component from the radar wave data By estimating the transition path on which the local periodic component related to the state transitions on the state transition model, and comparing the attribution state to each transition pattern of each partial path constituting the transition path and the pulse identification determination rule A pulse determination unit that identifies a species and outputs an identification result; and a display unit that presents to a user a source type related to a received radar wave or that identification is impossible based on the input identification result It is what I did.
[0010]
In the radar identification device according to the present invention, the pulse classification unit verifies a transition path of a local periodic component related to an arbitrary radar wave, and specifies each transition pattern to which each partial path constituting the transition path belongs. The state transition model is modified when the number of possible transmission source types is 0 or 2 or more.
[0011]
A radar identification device according to the present invention extracts one or a plurality of radio wave specifications from radar wave data, and identifies radar wave specifications for identifying a transmission source type related to a radar wave received based on the extracted radio wave specifications. A discriminator, and the display unit is capable of identifying the source type of the received radar wave based on the discrimination result output from the pulse determination unit and the discrimination result output from the radar wave specification discrimination unit This is what is presented to the user.
[0012]
A radar identification device according to the present invention includes a base station configured to include a pulse classification unit, and a portable device configured to include a radar wave reception unit, a pulse determination unit, and a display unit. It is.
[0013]
A radar identification device according to the present invention includes an automatic device configured with a radar wave receiving unit, a base station configured with a pulse classification unit, a pulse determination unit, and a display unit. Mobile phone.
[0014]
A radar identification device according to the present invention includes a base station configured to include a pulse classification unit, a portable unit configured to include a radar wave reception unit, a pulse determination unit, a radar wave specification identification unit, and a display unit. Are provided.
[0015]
A radar identification device according to the present invention includes an automatic device having a radar wave receiving unit, a base station having a pulse classification unit, a pulse determination unit, a radar wave specification identification unit, and a display A portable device having a portion.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described.
Embodiment 1 FIG.
FIG. 1 is a block diagram showing a configuration of a radar identification apparatus according to Embodiment 1 of the present invention. In FIG. 1, reference numeral 1 denotes a radar wave receiving unit that receives radar waves and generates radar wave data, and 2 generates pulse data related to a pulse waveform from radar wave data input from the radar wave receiving unit 1 and the pulse data. A pulse classification unit for creating a rule for identifying a transmission source type (hereinafter referred to as a pulse identification determination rule) based on a classification result for partial feature values extracted in time series from 3, a radar wave receiving unit 1. Input radar wave data from 1 and input a pulse identification judgment rule from pulse classification unit 2 to identify the source type related to the received radar wave and output the identification result (hereinafter referred to as pulse identification result). The pulse determination unit 4 receives radar wave data from the radar wave receiving unit 1 and identifies the source type based on radio wave specifications such as frequency, PRF, and pulse width. The radar wave specification identifying unit for identifying the source type related to the received radar wave based on the rule (hereinafter referred to as the specification identification determination rule) and outputting the identification result (hereinafter referred to as the specification identification result) 5 inputs a pulse identification result from the pulse determination unit 3 and a specification identification result from the radar wave item identification unit 4, and comprehensively identifies the source type from the pulse identification result and the item identification result. It is a display unit that identifies a transmission source type related to a radar wave received based on a rule (hereinafter referred to as a comprehensive identification determination rule) and presents the identification result to a user.
[0017]
Next, the operation of the radar identification apparatus according to the first embodiment of the present invention shown in FIG. 1 will be described. Here, some terms used in the following description of operations are defined in advance. A radar wave whose transmission source type related to the received radar wave is known in advance or by the function of the radar wave receiving unit 1 is referred to as “a radar wave with identification”. Further, a radar wave whose transmission source type relating to the received radar wave is unknown is referred to as an “unidentified radar wave”. Radar identification apparatus for inputting pulse data with identification and unidentified radar wave data based on radio wave reception etc., and performing unsupervised classification based on these radar wave data to construct a pulse identification determination rule The usage form is referred to as “identification preparation form”. Furthermore, a usage form of the radar identification device when receiving an unidentified radar wave and identifying a transmission source type related to the received radar wave is referred to as an “identification embodiment”.
[0018]
The radar wave receiving unit 1 receives radar waves and generates radar wave data. The generated radar wave data includes, for example, main data composed of time-series radar wave intensity values obtained by measuring radar wave intensity at predetermined time intervals, radar wave intensity measurement time intervals, and identification data. In the case of a radar wave, it is configured to have sub-data consisting of identification information (hereinafter referred to as a transmission source identifier) of a transmission source type related to the radar wave with identification. Therefore, the radar wave data related to the received radar wave is matched with the radar wave with identification in which the transmission source type is clear and the transmission source identifier is added, and the radar wave with identification and the transmission source type is unknown. Therefore, it is classified into unidentified radar wave data to which a temporary identifier is added. For the radar wave with identification, the radar wave receiving unit 1 outputs the radar wave data only to the pulse classification unit 2. For the unidentified radar wave, the radar wave receiving unit 1 outputs the radar wave data only to the pulse classification unit 2 in the identification preparation form, and in the identification embodiment, the radar wave data is output to the pulse classification unit 2 and the pulse determination unit 3. And output to the radar wave specification identifying unit 4.
[0019]
FIG. 2 is a flowchart showing the operation of the pulse classification unit. When receiving the radar wave data from the radar wave receiving unit 1, the pulse classifying unit 2 appropriately divides the radar wave data and takes out and stores the pulse data (step ST1). In addition, it is known that the pulse waveform identified by the pulse data has a unique shape characteristic with respect to each component member such as a magnetron constituting the radar wave source apparatus. That is, the characteristics for each transmission source type appear in the pulse waveform. Next, based on the extracted pulse data, the pulse classification unit 2 performs conversion by applying a predetermined calculation to the time-series data for identifying the pulse waveform, thereby performing a partial feature (hereinafter referred to as a partial feature amount). , Referred to as a local periodic component) in time series (step ST2). Next, for the obtained local periodic component, the pulse classifying unit 2 classifies the local periodic component into a predetermined number of categories based on the features, and a plurality of patterns each defining a feature amount representing the category. The classification according to which local pattern belongs to (hereinafter referred to as a local pattern) is performed, and the classification result is stored (step ST3). If the classification result is obtained, the pulse classification unit 2 detects the belonging state to each local pattern of the radar wave data with identification, and configures a pulse identification determination rule defined to match the belonging state, The configured pulse identification determination rule is output to the pulse determination unit 3 (step ST4). It is preferable that the pulse classification unit 2 has a display function and a control function so that the configuration of the pulse identification determination rule by the pulse classification unit 2 can be monitored and controlled and the pulse identification determination rule can be corrected. In addition, the pulse classification unit can efficiently create a pulse determination and identification rule by inputting the radar wave data with identification and the unidentified radar wave data created or accumulated not only in the radar wave receiving unit 1 but also in another device. 2 preferably has a data input interface function.
[0020]
Next, each process shown as said step ST1 to step ST4 is demonstrated in detail. In step ST1, a radar wave intensity value at each time is detected from the radar wave data, and a time zone in which the radar wave intensity value is larger than a predetermined threshold is determined as a pulse time. Further, radar wave data in a time zone obtained by adding a predetermined time before and after the pulse time is divided from other portions and extracted as pulse data, and the pulse data is stored in predetermined storage means. When the received radar wave is a radar wave with identification, a transmission source identifier is added as attribute data to each pulse data. When the received radar wave is an unidentified radar wave, a temporary identifier is added as attribute data to each pulse data.
[0021]
In step ST2, by applying a predetermined calculation to the pulse data generated in step ST1, conversion to local features is performed, time series data of local periodic components is calculated, and these local data are calculated. The time-series data of the periodic component is stored in a predetermined storage means. As an example of calculating the local periodic component, pulse data in which the radar wave intensity is measured S times within the pulse time will be described. The radar wave intensity values obtained by measurement at four consecutive times are arranged in order to generate a quaternary vector, and this vector generation is performed by shifting by two unit times. That is, (S-2) / 2 vectors can be obtained by performing such vector generation over the entire pulse data. Next, each vector V = (v ₁ , V ₂ , V ₃ , V ₄ ) For the local periodic component X = (x ₁ , X ₂ , X ₃ , X ₄ ) Is calculated. The transformation according to equation (1) is a simple Walsh transform, but a discrete Fourier transform or the like may be applied.
[Expression 1]

[0022]
In step ST3, T local periodic components (X) obtained from all the pulse data related to the input radar wave data. ₁ ~ X _T ) For each, unsupervised classification is performed to determine which local pattern belongs to a predetermined number (n) of local patterns without referring to an identifier. FIG. 3 is a flowchart showing the unsupervised classification operation related to the local periodic component. First, an initial value for setting each parameter defining a mixed normal distribution representing the whole of a plurality of local patterns, each given as a normal distribution, by the expected value maximization method shown below is set (substep ST11). For example, T local periodic components X given for each radar wave ₁ ~ X _T N different ones are appropriately selected to obtain an average initial value for each local pattern, and X ₁ ~ X _T The variance is calculated from the initial value of the variance common to n local patterns, and the initial value of the mixing ratio is 1 / n common to each of the n local patterns. A set of mean, variance, and mixture ratio (μ _i , Σ _i , Α _i ) (Hereinafter referred to as a classification parameter), the local periodic component (X ₁ ~ X _T ) For each local pattern probability density p ₁ ~ P _n Is calculated based on the following equation (2).
p _i = Α _i N (μ _i , Σ _i (2)
In Formula (2), i = 1 to n, and N (μ _i , Σ _i ) Indicates the probability density of the normal distribution. In addition, each local periodic component X _t Probability of belonging P to each local pattern i _ti Is calculated based on the following equation (3) (sub-step ST12).
[Expression 2]

Each local periodic component X _t Probability of attribution P _ti Is obtained from the following equation (4), the classification parameter (μ _i , Σ _i , Α _i ) Is calculated (substep ST13). In formula (4), the case where only the diagonal covariance component is calculated is taken as an example.
[Equation 3]

In addition, a classification parameter (μ _i , Σ _i , Α _i ) Is recorded, the history related to the transition of the classification parameter is referred to, the verification related to the history data based on a predetermined standard is performed, and the sub-step to be executed next is selected according to the verification result. For example, if the fluctuation range relating to the average, variance, and mixture ratio values does not exceed a predetermined threshold over the 10 classification parameter calculation steps executed so far, the data attribution maximum likelihood estimation sub-step Proceed to ST14, otherwise return to data attribution probability calculation sub-step ST12. Classification parameter consisting of mean, variance and mixing ratio (μ _i , Σ _i , Α _i If the convergence state is confirmed without the change relating to) exceeding the predetermined threshold, the classification parameter (μ _i , Σ _i , Α _i ) As a classification parameter defining a mixed normal distribution. Next, using the classification parameter, the same processing as in data attribution probability calculation sub-step ST12 is performed, and each local periodic component X _t For each local pattern i _ti Is calculated. And each local periodic component X _t For each (t = 1 to T), the attribution probability P _ti By comparing the maximum attribution probability P _ti The local periodic component X _t And the local periodic component X _t And the local periodic component X _t Is recorded in association with the local pattern to which the attribute belongs (substep ST14).
[0023]
In step ST4, for each local pattern i, the local periodic component X in which attribution to the local pattern i is recorded. _t The identifier added as attribute data is checked to determine whether or not the local pattern i can be a local pattern unique to a specific source type. That is, the local periodic component X in which the attribution to the local pattern i is recorded _t If the source identifier is added as attribute data, if the type of the source identifier is single, the local pattern i is a local pattern specific to the source type specified by the source identifier. judge. If a plurality of transmission source identifiers are mixed, the first transmission source type related to the transmission source identifier having the largest number of local periodic components in which attribution to the local pattern i is recorded is recorded. Compared with the second transmission source type related to the transmission source identifier having the second largest number of local periodic components, the number of recorded local periodic components for the first transmission source type is about the second transmission source type. If the number of recorded local periodic components exceeds the predetermined threshold, it is determined that the local pattern i is a local pattern unique to the first source type, and exceeds the predetermined threshold. If not, it is impossible to judge. As described above, for each local pattern i, an individual determination rule (hereinafter referred to as an individual determination rule) is obtained by associating the local pattern with a transmission source type that can be specified by the local pattern. If the existence of the individual decision rule is confirmed for all local patterns, the pulse identification decision rule is constructed by collecting the data obtained by combining the individual decision rule and the classification parameter obtained for each local pattern, and the pulse The identification determination rule is output to the pulse determination unit 3.
[0024]
Next, FIG. 4 is a flowchart showing the operation in the pulse determination unit. When the unidentified radar wave data is input from the radar wave receiving unit 1, the pulse determining unit 3 appropriately divides the radar wave data and extracts and stores the pulse data (step ST21). Next, the pulse determination unit 3 performs conversion by applying a predetermined calculation to the time series data related to the pulse waveform based on the extracted pulse data, thereby converting the local periodic component into the time series. (Step ST22). Next, the pulse determination unit 3 identifies the transmission source type related to the unidentified radar wave based on the attribution state of each local periodic component generated in step ST22 with reference to the pulse identification determination rule. And the said identification result, ie, a pulse identification result, is output to the display part 5 (step ST23).
[0025]
Since step ST21 is the same as step ST1 executed by the pulse classification unit 2, and step ST22 is the same as step ST2 executed by the pulse classification unit 2, detailed description thereof is omitted. In step ST23, each local periodic component X related to unidentified radar wave data is based on the classification parameter included in the pulse identification determination rule input from the pulse classification unit 2. _t Calculate the belonging probability to each local pattern i every time and calculate the largest belonging probability P _ti The local periodic component X _t Is assumed to belong. The calculation of the probability of attribution is similarly performed using equation (3). Next, based on the individual determination rules for each local pattern constituting the pulse identification determination rule, the local pattern associated with the transmission source type is detected from among the local patterns to which each local periodic component belongs. To do. Next, for each local pattern that is associated with the transmission source type, the number of local periodic components belonging to the local pattern is aggregated and is associated with the local pattern with the largest number of local periodic components. The transmission source type is determined to be a transmission source type related to the unidentified radar wave received, and the determination result is output to the display unit 5 as a pulse identification result. Also, if there is no local pattern associated with the transmission source type among the local patterns to which each local periodic component related to the received radar wave belongs, or is associated with the transmission source type. When the number of local periodic components belonging to the local pattern is smaller than a predetermined threshold value, a pulse identification result indicating that determination is impossible is output to the display unit 5.
[0026]
Next, the radar wave item identification unit 4 identifies a transmission source type related to unidentified radar wave data input from the radar wave reception unit 1 with reference to a given item identification determination rule, and the identification result That is, the specification identification result is output to the display unit 5. Note that, as described above, the radio wave specifications serving as a reference for identifying the transmission source type are the specification values such as frequency, PRF and pulse width, and the specification modes such as frequency mode, PRF mode and pulse mode. Since radar identification based on these radio wave specifications is a well-known technique as disclosed in, for example, Japanese Patent Application Laid-Open No. 11-326494, description thereof is omitted here.
[0027]
Next, the display unit 5 inputs the pulse identification result from the pulse determination unit 3 and also inputs the item identification result from the radar wave item identification unit 4. If these identification results are obtained, the source type related to the unidentified radar wave received from the pulse identification result and the specification identification result is identified with reference to the given comprehensive identification determination rule, and the identification result is determined by the user. To present. An example of the comprehensive identification determination rule will be described below. If the pulse identification result and the item identification result match, or if either the pulse identification result or the item identification result cannot be determined, it is specified based on the pulse identification result or the item identification result. The source type is displayed to the user. In addition, when both the pulse identification result and the specification identification result cannot be determined, it is displayed to the user that determination is impossible. In addition, when the pulse identification result and the specification identification result do not match, it is specified in advance which identification result is to be prioritized, and the transmission source type specified based on the priority identification result is determined by the user. Or the source type specified based on both identification results is displayed to the user. In addition, it is preferable that the display unit 5 has a control function for changing the display mode so that various display modes related to the transmission source type and the like in the display unit 5 can be selected.
[0028]
FIG. 5 is a block diagram showing a configuration of a modification of the radar identification device according to Embodiment 1 of the present invention. The radar identification device shown in FIG. 5 is different from the radar identification device shown in FIG. 1 in that the radar wave specification identification unit 4 is deleted. That is, based on only the pulse identification result output from the pulse determination unit 3, the display unit 5 presents to the user that the transmission source type related to the received radar wave or that identification is impossible.
[0029]
As described above, according to the first embodiment, the radar wave receiving unit 1 that generates radar wave data and the pulse classification unit that constitutes the pulse identification determination rule for identifying the source type based on the pulse data. 2 and a pulse determination unit 3 for identifying a transmission source type by comparing a state of belonging to each local pattern and a pulse identification determination rule for each local periodic component related to pulse data related to the received radar wave, and identification Since it is configured to include the display unit 5 that presents to the user, for example, a transmission source type related to the radar wave received based on the result, each component unique to the magnetron or the like that constitutes the radar wave source device Because the source type is identified based on the pulse waveform that appears, accurate identification results can be obtained even if errors occur in the settings related to radio wave specifications in the transmission and reception of radar waves. Getting possible, an effect that it is possible to improve the identification accuracy.
[0030]
In addition, since the pulse classification unit 2 configures the pulse identification determination rule based on the attribution state of each local periodic component obtained in time series for each radar wave to each local pattern, It is possible to use the unsupervised classification result related to the included local pattern, and it is possible to obtain an effective pulse identification judgment rule and ensure the desired identification accuracy even if there are few radar waves with identification for learning. There is an effect that practicality can be improved.
[0031]
In addition, the pulse classification unit 2 obtains the probability of belonging to each local pattern of each local periodic component based on a mixed normal distribution that represents the entire plurality of local patterns given as normal distributions, respectively, so that the local period Since the local pattern to which the component belongs is determined, the local pattern can be classified using a statistically effective method, so the reliability of the pulse identification judgment rule can be improved and transmitted. There is an effect that the identification accuracy of the source species can be improved.
[0032]
Also, a radio wave specification is extracted from the received radar wave, and a radar wave specification identification unit 4 for identifying a transmission source type based on the extracted radio wave specification is provided. Since the transmission source type related to the received radar wave is presented to the user based on the pulse determination result and the item identification result output from the radar wave item identification unit 4, In addition, since the source type can be identified based on various radio wave specifications, the identification accuracy can be further improved.
[0033]
Embodiment 2. FIG.
The radar identification device according to the second embodiment has the same configuration as that of the radar identification device according to the first embodiment. The operation related to the configuration of the pulse identification determination rule in the pulse classification unit, and the pulse determination This embodiment is different from the first embodiment in that the identification operation related to the unidentified radar wave in the unit is different.
[0034]
Hereinafter, the operation of the radar identification apparatus according to the second embodiment will be described. The operations related to the radar wave receiving unit 1, the radar wave specification identifying unit 4 and the display unit 5 are the same as those in the first embodiment, and thus the description thereof is omitted. FIG. 6 is a flowchart showing the operation of the pulse classification unit according to the radar identification apparatus according to Embodiment 2 of the present invention. When the pulse classification unit 2 inputs the radar wave data from the radar wave receiving unit 1, the pulse classification unit 2 applies a predetermined calculation to the radar wave data in the same manner as the pulse data in the first embodiment. Conversion to local feature values is performed, and local periodic components are extracted in time series (step ST31). Next, the pulse classification unit 2 includes a plurality of local patterns to which local periodic components given in time series for radar waves can belong, and a plurality of transition patterns that define transitions between local patterns (including autoregressive). Based on a model (hereinafter referred to as a state transition model) that expresses temporal changes in feature quantities related to radar waves, local periodic components obtained in time series for each radar wave are represented as feature quantity changes on the state transition model. Model parameter that defines the change state of the entire radar wave in the state transition model, as well as estimating the most likely transition route in the route traced according to the following (hereinafter referred to as transition route). (Step ST32). In addition, in the transition path given based on the change of the local periodic component given in time series for an arbitrary radar wave, a part of the transition path is set for the change of the local periodic component adjacent in time. A path to be configured is referred to as a partial path. In addition, each partial path belongs to a transition pattern defined between local patterns to which local periodic components that are temporally adjacent to each other are set. In addition, since the partial paths are set in time series as described above, it is naturally expected that a plurality of partial paths belong to the same transition pattern, and the same transition of such partial paths. The degree of overlap with the pattern is also an important element for expressing the belonging state to each transition pattern for each partial path related to the radar wave. Next, the pulse classification unit 2 performs pulse based on the belonging state to each transition pattern of each partial path set between adjacent local periodic components for the local periodic components obtained in time series for each radar wave with identification. An identification determination rule is generated (step ST33). It is preferable that the pulse classification unit 2 has a display function and a control function so that the configuration of the pulse identification determination rule by the pulse classification unit 2 can be monitored and controlled and the pulse identification determination rule can be corrected.
[0035]
Next, each process shown as said step ST31 to step ST33 is demonstrated in detail. In step ST31, as in the case of the pulse data in the first embodiment, a predetermined calculation is applied to the radar wave data input from the radar wave receiving unit 1 to obtain a local feature amount. The local periodic components are calculated in time series, and these local periodic components are stored in a predetermined storage means.
[0036]
In step ST32, as described above, model parameters are calculated based on time series data of local periodic components given for each radar wave based on a given state transition model. FIG. 7 is a diagram illustrating an example of a state transition model. In FIG. 7, A1 is an inter-pulse state, A2 is a pulse start state, A3 is a pulse state, A4 is a pulse end state, and Bi (i = 1 to 13) is a local pattern to which each local periodic component can belong. , R _i (I = 1 to 11) are transition patterns that define possible transitions between local patterns based on the characteristics of radar wave data. As shown in FIG. 7, this state transition model is given as a model assuming a loop-like state transition consisting of four states of an inter-pulse state A1, a pulse start state A2, a pulse state A3, and a pulse end state A4. Yes. Each state also has a transition pattern that is autoregressive and has an arbitrary number of local patterns. For example, the local patterns B3, B4, and B5 that the pulse start state A2 has form a fully coupled transition network with each other, and each has a transition pattern R2 from the interpulse state A1 and a transition pattern R7 to the pulse state A3. Have. Regarding the transition patterns R2 and R7, instead of setting one transition pattern for the local patterns B3, B4, and B5, three transition patterns are set for each of the local patterns B3, B4, and B5. May be. Moreover, each local pattern to which a local periodic component belongs can be expressed by, for example, a single normal distribution. Transitions between these local patterns are determined probabilistically, and the probabilities are referred to as transition probabilities. Further, for each local pattern, a parameter configured by combining a normal distribution parameter expressing the local pattern and a transition probability relating to each transition pattern extending from the local pattern is referred to as a model parameter.
[0037]
FIG. 8 is a flowchart showing a model parameter calculation process in the state transition model. Here, T local periodic components obtained in time series for each radar wave data are represented by X ₁ ~ X _T And the number of local patterns is n. First, initial values for obtaining model parameters by the expected value maximization method shown below are set (substep ST41). For n local patterns, for example, a local periodic component X given in time series for each radar wave ₁ ~ X _T N different ones are appropriately selected to obtain an average initial value for each local pattern, and X ₁ ~ X _T Then, the variance is calculated from the initial value of the variance common to the n local patterns, and the transition probability is the equal probability. That is, regarding the transition probability, if four transition patterns extend from an arbitrary local pattern, the transition probability related to each transition pattern is set to 0.25. Model parameters consisting of a set of mean, variance, and transition probability for each local pattern (μ _i , Σ _i , Α _i ) For each radar wave, the local periodic component X related to the radar wave. ₁ ~ X _T The most likely transition path for is detected (substep ST42). If a transition path related to each radar wave is detected, a model parameter (μ _i , Σ _i , Α _i ) Is calculated (substep ST43). When the model parameter is calculated in sub-step ST43, the history related to the transition of the model parameter is referred to, the verification related to the history data based on a predetermined standard is performed, and the next execution is performed according to the verification result Select the substep to be performed. For example, if the fluctuation range relating to the average, variance, and transition probability values over the 10 model parameter calculation steps executed so far exceeds a predetermined threshold value, go to the transition path maximum likelihood estimation sub-step ST42. Otherwise, model parameters (μ _i , Σ _i , Α _i ) Is in a convergent state and the model parameter (μ _i , Σ _i , Α _i ) As a model parameter for determining the state transition model.
[0038]
In step ST33, for each radar wave, the transition path is divided by omitting the transition between the local patterns included in the inter-pulse state A1, so that the transition path corresponding to the pulse waveform (hereinafter referred to as a pulse transition path). ). Next, for each radar wave with identification, the transmission source identifier is attached as attribute data to a transition pattern to which a partial path passing through the pulse transition path related to the radar wave with identification belongs. When the processing for attaching the transmission source identifier to the transition pattern is completed for all the radar waves with identification, for each transition pattern, the transmission source identifier in which the attachment to the transition pattern is recorded is examined to identify the transition pattern. A determination is made as to whether or not a transition pattern unique to the source type can be obtained. For an arbitrary transition pattern in which the attachment of a transmission source identifier is recorded, if the type of the recorded transmission source identifier is single, the transition pattern is a transition pattern specific to the transmission source type specified by the transmission source identifier. Is determined. If a plurality of transmission source identifiers are mixed, the first transmission source type related to the transmission source identifier having the largest number of records attached to the transition pattern and the transmission having the second largest number of recordings. When the number of records for the first source type is greater than the number of records for the second source type by comparison with the second source type associated with the source identifier, It is determined that the transition pattern is a transition pattern unique to the first transmission source type, and determination is impossible if the predetermined threshold value is not exceeded. As described above, for each transition pattern, an individual determination rule (hereinafter referred to as an individual determination rule) is obtained by associating the transition pattern with a transmission source type that can be specified by the transition pattern. If the existence of individual judgment rules for all transition patterns is confirmed, a pulse identification judgment rule is constructed by collecting the data obtained by combining the individual judgment rules and model parameters obtained for each transition pattern. The identification determination rule is output to the pulse determination unit 3.
[0039]
Next, the operation of the pulse determination unit 3 will be described. The pulse determination unit 3 receives the radar wave data related to the unidentified radar wave from the radar wave receiving unit 1 and uses the pulse classification unit 2 for the time series data related to the radar waveform based on the radar wave data. Conversion is performed by applying a predetermined calculation similar to that performed, and time-series data of local periodic components is calculated. Next, the pulse determination unit 3 determines model parameters (μ) consisting of a set of average, variance, and transition probability determined for each local pattern. _i , Σ _i , Α _i ) Based on the time series data X of the local periodic component related to the radar wave data ₁ ~ X _T The most likely transition path for is detected. If the transition route with the highest likelihood is estimated, if the transition pattern to which the partial route passing through the transition route belongs is associated with the source type, the source identifier is registered. To do. When there is one registered source identifier, it is determined that the source type is related to an unidentified radar wave received with the source type specified by the source identifier, and the pulse identification result is displayed on the display unit 5. Output to. Further, when there are a plurality of registered source identifiers, it is determined that the source type is related to an unidentified radar wave received with the source type specified by the source identifier having the largest number of associated transition patterns. Then, the pulse identification result is output to the display unit 5. In addition, when there is no registered source identifier, or when there are a plurality of registered source identifiers and the number of transition patterns associated with any source identifier is less than a predetermined threshold, determination is impossible. Is output to the display unit 5.
[0040]
As described above, according to the second embodiment, the transmission source based on the radar wave receiving unit 1 that generates the radar wave data and the transition path on the state transition model for the local periodic component related to the radar wave data A pulse classification unit 2 constituting a pulse identification determination rule for identifying a species, a pulse determination unit 3 for identifying a source type by comparing a transition path estimated for a received radar wave and a pulse identification determination rule, The display unit 5 presents the transmission source type related to the radar wave received based on the identification result to the user. Therefore, each component member such as a magnetron that constitutes the radar wave source device is unique. The source type can be identified based on the pulse waveform that reveals its characteristics, so it is accurate even if there are errors in the settings related to radio wave specifications in the transmission and reception of radar waves. It can be obtained identification result, an effect that it is possible to improve the identification accuracy.
[0041]
Further, by providing the radar wave specification identifying unit 4, the same effect as described in the first embodiment can be obtained. Furthermore, because the source type is identified based on the transition path on the state transition model for the local periodic component, a set of local periodic components corresponding to the pulse waveform can be easily identified, and the pulse from the radar wave data can be identified. Preliminary work such as setting a reference value for extracting features related to the waveform can be omitted, and the user can be reduced in load and the usability can be improved.
[0042]
Embodiment 3 FIG.
The radar identification device according to the third embodiment has the same configuration as that of the radar identification device according to the second embodiment, except that the operation related to the configuration of the pulse identification determination rule in the pulse classification unit is different. This is different from the second embodiment.
[0043]
Hereinafter, the operation of the radar identification apparatus according to the third embodiment will be described. The operations relating to the radar wave receiving unit 1, the pulse determining unit 3, the radar wave specification identifying unit 4 and the display unit 5 are the same as those in the second embodiment, and thus description thereof is omitted. FIG. 9 is a flowchart showing the operation of the pulse classification unit according to the radar identification apparatus according to Embodiment 3 of the present invention. When the radar wave data is input from the radar wave receiving unit 1, the pulse classifying unit 2 applies a predetermined calculation to the time-series data related to the radar wave as in the second embodiment, thereby applying a predetermined calculation to the local feature amount. Conversion is performed to calculate time-series data of local periodic components (step ST51). Next, based on the state transition model, the pulse classification unit 2 estimates a transition path with the highest likelihood for the local periodic component obtained in time series for each radar wave, and relates to the entire radar wave in the state transition model. A model parameter that defines the change state is calculated (step ST52). Next, the pulse classification unit 2 based on the belonging state to each transition pattern of each partial path set between adjacent local periodic components for local periodic components obtained in time series for each radar wave with identification, If a predetermined condition is satisfied, a pulse identification determination rule is generated (step ST53). If the predetermined condition is not satisfied, the state transition model is deformed in order to improve the modeling accuracy with respect to the temporal change of the feature amount related to the radar wave (step ST54). It is preferable that the pulse classification unit 2 has a display function and a control function so that the configuration of the pulse identification determination rule by the pulse classification unit 2 can be monitored and controlled and the pulse identification determination rule can be corrected.
[0044]
Next, each process shown as said step ST51 to step ST54 is demonstrated in detail. Since step ST51 is the same as step ST31 in the second embodiment, the description thereof is omitted. Further, step ST52 is different from step ST32 in the second embodiment only in that the form of the state transition model to be set is changed, and the basic operation is the same as step ST32, so that the description thereof is omitted. . That is, the model parameter may be calculated based on a state transition model set as an initial model in advance, or the model parameter may be calculated based on the state transition model deformed by the state transition model deformation step ST54.
[0045]
In step ST53, the pulse transition path corresponding to the pulse waveform is extracted by dividing the transition path by omitting the transition between the local patterns included in the inter-pulse state A1 for each radar wave. Next, for each identified radar wave, the transmission source identifier associated with the identified radar wave is used as attribute data for the transition pattern to which each partial path that has passed through the pulse transition path associated with the identified radar wave belongs. Come with it. When the processing for attaching the transmission source identifier to the transition pattern is completed for all the radar waves with identification, for each transition pattern, the transmission source identifier in which the attachment to the transition pattern is recorded is examined to identify the transition pattern. A determination is made as to whether or not a transition pattern unique to the source type can be generated, and an individual determination rule is generated. Note that the determination process related to whether or not it can be a transition pattern unique to a specific transmission source type has already been described in step ST33 of the second embodiment, and thus description thereof is omitted here. Then, if the existence of the individual determination rule is confirmed for all transition patterns, for each pulse transition path, transitions to which the respective partial paths that have passed when following the pulse transition path belong for a predetermined number of pulse transition paths Counts the type of source identifier recorded for attachment to the pattern. When the number of types of transmission source identifiers, that is, the number of types of transmission source types is 0 or 2 or more, or when the number of executions of the determination rule determination step ST53 is equal to or less than a specified number, the state transition model transformation step ST54 is executed. To do. Further, when the number of types of transmission source types is 1, or when the number of executions of the determination rule determination step ST53 exceeds the specified number, data obtained by combining individual determination rules and model parameters for each transition pattern Are combined to form a pulse identification determination rule, and the pulse identification determination rule is output to the pulse determination unit 3.
[0046]
In step ST54, a model parameter is searched for each local pattern to detect a local pattern having the largest variance value, and the local pattern is divided to generate two new local patterns. The transition between the two local patterns is a full connection. In addition, for the initial values of model parameters related to each local pattern, for example, one local pattern is given the model parameter of the local pattern before division, and the other local pattern is generated up to a given minute amount. The model parameter obtained by correcting the model parameter of the local pattern before the division based on the random number is given.
[0047]
As described above, according to the third embodiment, the same effect as that of the second embodiment is obtained, and the pulse classification unit 2 verifies a transition path related to an arbitrary radar wave and configures the transition path. Since the state transition model is modified when the number of types of source types that can be specified by each transition pattern to which each partial path belongs is 0 or 2 or more, optimal state transition is automatically performed Since the model can be configured, there is no need to perform accurate modeling in advance, and the user's burden can be reduced.
[0048]
Embodiment 4 FIG.
FIG. 10 is a block diagram showing a configuration of a radar identification apparatus according to Embodiment 4 of the present invention. The radar identification device according to the fourth embodiment is different from the first to third embodiments in that the portable device is configured separately from the functional elements used in the identification embodiment. In FIG. 10, the same reference numerals as those in FIG. Reference numeral 11 is a base station having a pulse classification unit 2, and 12 is a portable device having a radar wave receiving unit 1, a pulse determining unit 3, a radar wave specification identifying unit 4 and a display unit 5. is there.
[0049]
Next, the operation will be described. Since the basic operation is the same as that of the radar identification device according to either the first embodiment or the third embodiment, only the characteristic operation of the radar identification device according to the fourth embodiment will be described here. To do. In the identification preparation form, the radar receiver 1 in one or a plurality of portable devices 12 has a radar wave related to the received radar wave when the communication line between the base station 11 and the portable device 12 is established. When data is transmitted to the pulse classification unit 2 in the base station 11 and the communication line between the base station 11 and the portable device 12 is not established, the radar wave data is recorded and the communication line is established The radar wave data recorded in (1) is transmitted to the pulse classification unit 2 in the base station 11. The pulse classification unit 2 configures a pulse identification determination rule based on a plurality of accumulated pulse data, and the pulse identification determination rule is set while the communication line between the base station 11 and the portable device 12 is established. It transmits to the pulse determination part 3 in the 1 or several portable device 12. FIG. In the identification embodiment, the operations related to the radar wave receiving unit 1 and the pulse classification unit 2 are the same as described above, and each portable device 12 communicates between the portable device 12 and the base station 11. Regardless of whether or not a line is established, each radar wave is individually received and the source type related to the radar wave is identified.
[0050]
As described above, according to the fourth embodiment, the base station 11 that has the same effect as any of the first to third embodiments and has the pulse classification unit 2, the radar, Since it comprises the portable device 12 comprised including the wave receiving unit 1, the pulse determining unit 3, the radar wave specification identifying unit 4 and the display unit 5, the identification device according to the identification embodiment is provided with only the portable device 12. Since the use can be realized independently, it is possible to reduce the size and weight of the identification device for identifying the transmission source type related to the unidentified radar wave. In addition, since the base station 11 can be shared by a plurality of portable devices 12, the cost of the entire system for performing radar identification can be reduced.
[0051]
When the radar identification device has a configuration that does not include the radar wave specification identification unit 4 as shown in the modification of the first embodiment, a base station 11 configured with the pulse classification unit 2; By adopting a configuration including the portable device 12 configured to include the radar wave receiving unit 1, the pulse determining unit 3, and the display unit 5, the same effects as described above can be obtained.
[0052]
Embodiment 5. FIG.
FIG. 11 is a block diagram showing a configuration of a radar identification apparatus according to Embodiment 5 of the present invention. The radar identification apparatus according to the fifth embodiment is different from the fourth embodiment in that an automatic device capable of autonomous operation is provided with a radar wave receiving unit. In FIG. 11, the same reference numerals as those in FIG. 21 is an automatic device that is configured to include the radar wave receiving unit 1 and is preferably capable of autonomous operation, 22 is a base station configured to include the pulse classification unit 2, 23 is a pulse determination unit 3, and radar wave specifications. It is a portable device that includes the identification unit 4 and the display unit 5. As the automatic device 21, for example, an aircraft flying by wireless operation or automatic control such as a small radio control helicopter is assumed.
[0053]
Next, the operation will be described. Since the basic operation is the same as that of the radar identification device according to any one of the first to third embodiments, only the characteristic operation of the radar identification device according to the fifth embodiment will be described here. To do. In the identification preparation form, the radar wave receiving unit 1 in the automatic device 21 bases the radar wave data on the received radar wave on the basis of the communication line between the automatic device 21 and the base station 22 established. If the communication line between the automatic device 21 and the base station 22 is not established, the radar wave data is recorded and recorded when the communication line is established. The received radar wave data is transmitted to the pulse classification unit 2 in the base station 22. The pulse classification unit 2 configures a pulse identification determination rule based on a plurality of accumulated pulse data, and the pulse identification determination rule is set while the communication line between the base station 22 and the portable device 23 is established. It transmits to the pulse determination part 3 in the 1 or several portable device 23. FIG. In the identification embodiment, the operations related to the radar wave receiving unit 1 and the pulse classification unit 2 are the same as described above, and each portable device 23 is a communication line between the automatic device 21 and the portable device 23. Are established, the radar waves are individually received to identify the source type related to the radar waves.
[0054]
As described above, according to the fifth embodiment, the same effect as any of the first to third embodiments can be obtained, and the radar wave receiving unit 1 is configured and preferably autonomously operable. A certain automatic device 21, a base station 22 configured with the pulse classification unit 2, and a portable device 23 configured with the pulse determination unit 3, the radar wave specification identification unit 4, and the display unit 5. Since the radar wave data related to the radar wave received from the automatic device 21 is transmitted, the use in the identification embodiment can be realized independently only by the portable device 23. There is an effect that the identification device for identifying the source type related to the radar wave can be reduced in size and weight. In addition, since the automatic device 21 and the base station 22 can be shared for a plurality of portable devices 23, the cost of the entire system for performing radar identification can be reduced.
[0055]
In the case where the radar identification device has a configuration that does not include the radar wave specification identification unit 4 as shown in the modification of the first embodiment, an automatic device 21 that includes the radar wave reception unit 1 and By providing a base station 22 configured with the pulse classification unit 2 and a portable device 23 configured with the pulse determination unit 3 and the display unit 5, the same effects as described above can be obtained. Can play.
[0056]
In addition, it should be noted that the radar identification device disclosed in the first to fifth embodiments described above is an example of the present invention and does not limit the technical scope thereof. The technical scope of the present invention is determined based on the description of the scope of claims, and technical ideas equivalent to the scope of claims and various design changes within the scope of the claims are included in the present invention. .
[0057]
【The invention's effect】
As described above, according to the present invention, a radar wave receiving unit that receives radar waves and generates radar wave data, and generates pulse data and extracts a local periodic component related to the pulse data. Each local period component obtained for each pulse data is attributed to any one of a plurality of local patterns classified into a predetermined number, and the local period related to the radar wave whose source type is known in advance The pulse classification unit that composes the pulse identification judgment rule based on the attribution state of each component to each local pattern, and the attribution status of each local periodic component related to the received radar wave to the local pattern and the pulse judgment identification rule A pulse determination unit for identifying the transmission source type and a display unit for presenting the transmission source type related to the received radar wave to the user. Therefore, since the source type is identified on the basis of the pulse waveform that reveals the unique characteristics of each component member such as the magnetron that constitutes the radar wave source device, there is an error in the settings related to the radio wave specifications in the transmission and reception of radar waves. Even if this occurs, an accurate identification result can be obtained, and the identification accuracy can be improved. In addition, since the pulse classification unit configures the pulse identification determination rule based on the attribution state of each local periodic component obtained in time series for each radar wave to each local pattern, including unidentified radar waves Since the unsupervised classification result relating to the local pattern can be used, it is possible to obtain an effective pulse identification determination rule even if there are few radar waves with identification for learning, and to ensure a desired identification accuracy, There is an effect that practicality can be improved.
[0058]
According to the present invention, the pulse classification unit obtains the probability of belonging to each local pattern of each local periodic component based on a mixed normal distribution that represents the entire plurality of local patterns each given as a normal distribution. Since the local pattern to which the local periodic component belongs is determined, the local pattern can be classified using a statistically effective method, and the reliability of the pulse identification determination rule can be improved. Therefore, there is an effect that the identification accuracy of the transmission source type can be improved.
[0059]
According to the present invention, a radar wave receiving unit that receives a radar wave and generates radar wave data, extracts a local periodic component related to the radar wave data, and extracts the radar wave for each radar wave based on a state transition model. Estimate the transition path where the local periodic component related to the transition on the state transition model, and the attribution state to each transition pattern for each of the partial paths that constitute the transition path of the radar wave whose source type is known in advance The pulse classification unit that constitutes the pulse identification determination rule based on the received signal is compared with the attribution state of each partial path constituting the transition path estimated for the received radar wave and the pulse identification determination rule A pulse determination unit for identifying the transmission source type and a display unit for presenting the transmission source type related to the received radar wave to the user. Since the source type is identified on the basis of the pulse waveform that reveals the unique characteristics of each structural member such as the magnetron that constitutes the device, even if errors occur in the settings related to the radio wave specifications in the transmission and reception of radar waves An accurate identification result can be obtained, and the identification accuracy can be improved. Also, since the source type is identified based on the transition path related to the local periodic component of the radar wave, a set of local periodic components corresponding to the pulse waveform can be easily identified, and the pulse waveform is related to the pulse waveform from the radar wave data. Preliminary work such as setting a reference value for extracting features can be omitted, and the user's burden can be reduced and usability can be improved.
[0060]
According to this invention, the pulse classification unit verifies the transition path of the local periodic component related to an arbitrary radar wave, and can be identified by each transition pattern to which the individual partial paths constituting the transition path belong. Since the state transition model is configured to be deformed when the number of types of source species is 0 or 2 or more, it is possible to automatically configure an optimal state transition model, and modeling with high accuracy in advance is possible. There is no need to carry out, and it is possible to reduce the burden on the user and improve usability.
[0061]
According to the present invention, the radar wave specification identifying unit that extracts one or a plurality of radio wave specifications from the radar wave data and identifies a transmission source type related to the radar wave received based on the extracted radio wave specifications. And the display unit presents the transmission source type related to the received radar wave to the user based on the identification result output from the pulse determination unit and the identification result output from the radar wave item identification unit. Since it comprised, since a transmission source kind can be identified not only based on not only a pulse waveform but various radio wave specifications, there exists an effect that identification accuracy can be improved more.
[0062]
According to this invention, since it is configured to include a base station configured to include a pulse classification unit and a portable device configured to include at least a radar wave receiving unit, a pulse determination unit, and a display unit, Since the use in the identification embodiment can be independently realized only with the portable device, an effect of reducing the size and weight of the device for identifying the transmission source type related to the unidentified radar wave can be achieved. In addition, since the base station can be shared for a plurality of portable devices, the cost of the entire system for performing radar identification can be reduced.
[0063]
According to the present invention, an automatic device configured with a radar wave receiving unit, a base station configured with a pulse classification unit, and a portable device configured with at least a pulse determination unit and a display unit. If the radar wave data related to the radar wave received from the automatic device is transmitted, the use in the identification embodiment can be realized independently only by the portable device. There is an effect that it is possible to reduce the size and weight of the device for identifying the source type related to the identification radar wave. In addition, since the automatic device and the base station can be shared for a plurality of portable devices, the cost of the entire system for performing radar identification can be reduced.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of a radar identification apparatus according to Embodiment 1 of the present invention.
FIG. 2 is a flowchart showing an operation related to a pulse classification unit in the radar identification apparatus according to Embodiment 1 of the present invention;
FIG. 3 is a flowchart showing an unsupervised classification operation.
FIG. 4 is a flowchart showing an operation related to a pulse determination unit in the radar identification apparatus according to Embodiment 1 of the present invention;
FIG. 5 is a flowchart showing a configuration of a modification of the radar identification device according to Embodiment 1 of the present invention;
FIG. 6 is a flowchart showing an operation related to a pulse classification unit in a radar identification device according to Embodiment 2 of the present invention;
FIG. 7 is a diagram showing an example of a state transition model used in a radar identification device according to Embodiment 2 of the present invention.
FIG. 8 is a flowchart showing a model parameter calculation process.
FIG. 9 is a flowchart showing an operation related to a pulse classification unit in a radar identification device according to Embodiment 3 of the present invention;
FIG. 10 is a block diagram showing a configuration of a radar identification apparatus according to Embodiment 4 of the present invention.
FIG. 11 is a block diagram showing a configuration of a radar identification apparatus according to Embodiment 5 of the present invention.
FIG. 12 is a block diagram showing a configuration of a conventional radar identification device.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Radar wave receiving part, 2 Pulse classification part, 3 Pulse determination part, 4 Radar wave item identification part, 5 Display part, 11,22 Base station, 12,23 Portable device, 21 Automatic machine, A1 Interpulse state, A2 Pulse start state, A3 pulse state, A4 pulse end state, B1-B13 local pattern, R1-R11 transition pattern.

Claims (9)

A radar wave receiver for receiving radar waves and generating radar wave data;
Generates pulse data by appropriately dividing radar wave data, extracts local periodic components given in time series as local feature values from the pulse data, and obtains each local periodic component obtained for pulse data for each radar wave Is attributed to any one of a plurality of local patterns classified into a predetermined number, and based on the belonging state to each local pattern of a local periodic component related to a radar wave whose source type is known in advance In order to identify a transmission source type based on pulse data by associating a transmission source type with a local pattern that can identify the transmission source type and assembling a combination of the corresponding transmission source type and the local pattern A pulse classification part constituting the pulse identification judgment rule of
Pulse data is generated by appropriately dividing the radar wave data, local periodic components are extracted from the pulse data, and each local periodic component is one of the same local patterns as set by the pulse classification unit. A pulse determination unit that identifies the source type and outputs the identification result by comparing the state of attribution of the local periodic component related to the received radar wave to each local pattern and the pulse identification determination rule, belonging to the local pattern When,
A radar identification device comprising: a display unit that presents to a user, for example, a transmission source type related to a received radar wave or an inability to identify based on an input identification result. Based on the mixed normal distribution that expresses the whole of a plurality of local patterns given as normal distributions, the pulse classification unit determines the attribution probability of each local periodic component to each local pattern. The radar identification device according to claim 1, wherein a local pattern to be determined is determined. A radar wave receiver for receiving radar waves and generating radar wave data;
Extracts local periodic components given in time series as local feature values from radar wave data, and defines multiple local patterns to which local periodic components given in time series can belong, and transitions between local patterns. Based on a state transition model consisting of a plurality of transition patterns, a radar is used to estimate the transition path where the local periodic component related to the radar wave transitions on the state transition model for each radar wave, and the source type is known in advance. A transmission source type and a transition pattern that can identify the transmission source type are associated with each other based on the belonging state to each transition pattern for each partial path that constitutes the wave transition path, and the associated transmission source type Pulses that compose pulse identification judgment rules for identifying source types based on radar wave data by assembling combinations with transition patterns And a classification unit,
Extracting local periodic components from radar wave data, estimating the transition path on which the local periodic component related to the radar wave received based on the state transition model transitions on the state transition model, and configuring each transition path A pulse determination unit that identifies the source type by comparing the attribution state to each transition pattern of the path and the pulse identification determination rule, and outputs the identification result;
A radar identification device comprising: a display unit that presents to a user, for example, a transmission source type related to a received radar wave or an inability to identify based on an input identification result. The pulse classification unit verifies the transition path of the local periodic component related to an arbitrary radar wave, and the number of types of transmission source types that can be specified by each transition pattern to which each partial path constituting the transition path belongs is The radar identification device according to claim 3, wherein the state transition model is deformed when the number is 0 or 2 or more. A radar wave specification identifying unit for extracting one or more radio wave specifications from the radar wave data and identifying a transmission source type related to the radar wave received based on the extracted radio wave specifications;
Based on the identification result output from the pulse determination unit and the identification result output from the radar wave specification identification unit, the display unit indicates that the source type related to the received radar wave is indistinguishable, etc. The radar identification device according to claim 1, wherein the radar identification device is presented to a user. A base station configured to include a pulse classification unit, and a portable device configured to include a radar wave receiving unit, a pulse determination unit, and a display unit. The radar identification device described. An automatic device configured with a radar wave receiving unit, a base station configured with a pulse classification unit, and a portable device configured with a pulse determination unit and a display unit The radar identification device according to claim 1 or 3. A base station configured to include a pulse classification unit, and a portable device configured to include a radar wave receiving unit, a pulse determining unit, a radar wave specification identifying unit, and a display unit. Item 6. The radar identification device according to Item 5. An automatic device configured with a radar wave receiving unit, a base station configured with a pulse classifying unit, a portable unit configured with a pulse determining unit, a radar wave specification identifying unit, and a display unit The radar identification device according to claim 5, further comprising a detector.

Images