201207654 六、發明說明： 【發明所屬之技術領域】 本發明係關於-種電力監測方法，特別是關於___種可做為預 庭各電器設備使賴耗電祕鮮之建立迴路電力資訊 【先前技術】 如今全球環保與節能意識高漲，節能減碳的綠能產業為相當熱 的項目’以居家用電來說，根據2008年美國用電資訊統計，居家用 消耗量佔總用電量的37〇/〇,因此’唯有了解用習慣用電行為，才 • 有效達到居家節能的目的。為了準確分析家庭用電行為，現今最常採 用於各電器上安裝電力計以量測出電器用電狀況。 然而’多個電力計的加裝係相當耗費祕成本，且衫的電力計 更將消耗更多_電量。因此，本發明係、針對上述困擾提出—種利用 迴路型電力計監灌庭用電資訊的方法，_建立包含電雜用順序 性及相關性之迴路電力資訊_，並且可藉由此迴路電力資訊模型辨 識出迴路上電器_電狀態，讓制者及時了解目訊狀態的 統計數據’並且可藉由分析用電習慣數據，提供準確的省電建議。 【發明内容】 _ 本發明之主要目的係在提供-種建立迴路電力資訊模型之方法， 其係透過建立迴路電器使用狀態機率模型將電器使㈣順序性與相關 性量化’有效提升迴路上電器狀態組合的辨識程度。 本發明之;)¾ _目的係在提供-種建立迴路電力資訊模型之方法， 其係制型電力計進行制，並可設置於配錢巾管理將 大幅減少系統成本以及耗電量。 本發明之再-目的係在提供-種建立迴路電力資訊模型之方法， 其係能夠«與智雜電表結合，且可與各⑼力統計法搭配進 行即時監控電力’以分析統計出使用者細部用電，將大幅提能的 目楹。 3 201207654 為達到上述之目的，本發明提出之建立迴路電 法，其係其包含步驟為利用迴路雷 、 資訊與迴路上之複數個電料心 座電錶收集迴路之迴路用電 、電盗狀態資訊，且將依序擷取迴路用電 資訊之複數個時間區段的實功率與乏功率，藉由實功率與乏功^ 相對應每一時間區段之特徵向量 /、 訓練資料，經由計算每-機率模型訓練 之電器使^狀^隸由時間相計算出相對應該機率_訓練資料 吏用狀I、機率值，以建立迴路電器使用狀態機率模型。 發明貫施例配合所附的圖式詳加說明’當更容易瞭解本 發月之目的、技術内容、特點及其所達成之功效。 【實施方式】 本發明提ίϋ-種建立迴路電力資訊觀之方法其铜時考量同 -迴路上複數個電器使㈣順序性與相關性，從迴路所收集之迴 路用電資訊與電ϋ狀態資訊建立機率麵訓練資料，再經由計算機率 模型訓練料之_相，計算出姆賴賴到練將之電器使 用，態機率值，以建立-财fll使紐腿率_。底下則將以較 佳實施例詳述本發明之技術特徵。 第圖為本發明之建立迴路電力資訊模塑之流程圖，如圖所示， 首先’如，S1G ’於至少—迴路上綱至少一迴路紐簡定取樣 鮮取得此迴路上之電壓、功率目？及視在功率做為迴剌電資訊。 並且此迴，上之複數個電器係利用複數個插座電錶，以固定時間範 ，’量取每-電器之運糊電做為成為電綠態資訊，且經由計算電 器狀態資訊，求得各電H之各織所雜之最大視在轉與最小視在 功率’並依據電^之運作用電標記出狀態資訊之複數個狀態分切 點，以區分各電器之不同運作狀態。 之後如步驟S12，將利用滑動視窗方式，依序梅取迴路用電資 訊之複數個時間H段的實功率與乏功率，再依據所取得之實功率與乏 201207654 功率计算相對應母一時間區段之一特徵向量，每一特徵向量係包含平 均值、均方根值、變異數、最大值、最小值、最大差值及波形因子等 ，數個特，值，域形因子係為藉由最大值、平均值與均方根值所計 算出之波高率(Ccest Facto)、波形雜(F_ FaetQ「）、峰值功率比 (Peak to Average Ratio)及小波轉換係數(Wavelet transf〇rm)。 最後’如步驟S14 ’依據_區聽合電錄態f軸特徵向量 為複數個機率模型訓練資料，且經由計算每—機率模型訓練資料之電 器狀態資贿特徵向量’產生相對應解模型訓練資料之—電器特徵201207654 VI. Description of the invention: [Technical field to which the invention pertains] The present invention relates to a method for monitoring power, in particular, regarding the establishment of loop power information for the ___ kinds of electrical equipment in the court. Previous technology] Today's global awareness of environmental protection and energy conservation is high, and the green energy industry that saves energy and reduces carbon is a very hot project. In terms of household electricity, according to the 2008 US electricity consumption statistics, household consumption accounts for total electricity consumption. 37〇/〇, so 'only understand the use of customary electricity, only to effectively achieve the purpose of home energy saving. In order to accurately analyze the behavior of household electricity, it is most commonly used to install electric meters on various electrical appliances to measure the electrical condition of electrical appliances. However, the installation of multiple power meters is quite costly, and the power meter of the shirt will consume more power. Therefore, the present invention proposes a method for utilizing a loop type electric power meter to monitor the power consumption information of the hospital, and establishes a loop power information _ including the sequence and correlation of the electric miscellaneous, and can thereby generate power through the loop The information model identifies the electrical status of the electrical circuit on the loop, allowing the system to keep abreast of the statistical data of the visual state' and can provide accurate power saving advice by analyzing the electrical habit data. SUMMARY OF THE INVENTION The main object of the present invention is to provide a method for establishing a loop power information model, which is to quantify the order and correlation of the electrical appliances by establishing a circuit state probability model of the circuit appliance to effectively improve the electrical state of the circuit. The degree of recognition of the combination. The invention is directed to providing a method for establishing a loop power information model, which is implemented by a system type power meter and can be disposed in the distribution of the money towel to substantially reduce system cost and power consumption. A further object of the present invention is to provide a method for establishing a loop power information model, which can be combined with a smart meter and can be used to monitor power in real time with each (9) force statistics method to analyze and count user details. The use of electricity will greatly increase the visibility of the project. 3 201207654 In order to achieve the above object, the present invention proposes to establish a loop electrical method, which comprises the steps of utilizing loop lightning, information and a plurality of electrical material cores on the circuit to collect circuit loop power and electronic thief status information. And will sequentially capture the real power and the lack of power of the plurality of time segments of the loop power information, by calculating the eigenvectors/training data of each time segment corresponding to the real power and the spent power ^, by calculating each - The probability model training of the electrical appliance makes the shape of the ^ phase ^ from the time phase to calculate the relative probability _ training data 吏 use I, probability value, to establish a loop electrical appliance use state probability model. The invention is described in detail in conjunction with the accompanying drawings. It is easier to understand the purpose, technical content, characteristics and effects of this month. [Embodiment] The present invention provides a method for establishing a loop power information view. The copper time considers the plurality of electrical appliances on the same circuit to make (4) the sequence and correlation, and the loop power information and power state information collected from the loop. Establish the probability face training data, and then calculate the _ phase of the training material through the computer rate model, calculate the use of the electric machine, and the value of the state rate to establish the profit rate. The technical features of the present invention will be described in detail below with reference to preferred embodiments. The figure is a flow chart of the circuit power information molding of the present invention. As shown in the figure, first, for example, S1G 'at at least the loop is at least one loop, and the voltage and power are obtained on the loop. ? And the apparent power is used as the information. And this time, a number of electrical appliances on the system use a plurality of socket meters to fix the time range, 'measuring each of the electrical pastes as the electrical green state information, and calculating the electrical status information to obtain the electricity The maximum apparent turn-around and minimum apparent power of each of H's weaving is based on a plurality of state cut-off points of the state information according to the operation of the electricity to distinguish the different operating states of the electrical appliances. Then, in step S12, the sliding window mode is used to sequentially take the real power and the spent power of the H segment of the loop power information, and then calculate the corresponding time zone according to the obtained real power and the lack of 201207654 power. One of the eigenvectors of the segment, each eigenvector system includes an average value, a root mean square value, a variance number, a maximum value, a minimum value, a maximum difference value, and a waveform factor, etc., and a number of special values, domain shape factors are The Chest Facto, the waveform (F_ FaetQ), the Peak to Average Ratio, and the Wavelet Transf〇rm are calculated from the maximum, average, and rms values. 'If step S14' is based on the _ zone listening to the recorded state f-axis feature vector as a plurality of probability model training materials, and by calculating the electrical state status bribe feature vector of each probability model training data to generate corresponding solution model training data - electrical characteristics
方程式、-«狀態順序特徵方程式及—電驗_依特徵方程式， 以表示機轉_練㈣之咖細，並且再藉由咖相計算出機 率模型訓練資料之電ϋ使驗態機率值，以建立—迴路電器使 機率模型。 & 以上為本發贼立迴路電力資賴伽整體步職明底下將以 匕路上，、有棱燈電風扇以及一電鋼為範例，對於步驟引4整 合電器狀態資訊與特徵向量建立_路電^制織齡模型做 步詳細說明，請參閱第二圖、第三圖所示。 在上述步驟S14中，將依據時間區段整合電驗騎訊與特徵向 量為複數個機賴__料，請參_二圖本發明之機顿型訓練 資料之不意@，如圖所示’每-個機率模型訓練資料彳㈣域燈 風扇、電鋼之電器狀癌資訊及相對應之特徵向量(圖中未示)組成，並且 檯燈、電風扇、電齡電器在機率模型訓練資料1〇中所估之多數 狀態，將代表於此機率模型訓練資料1〇中之主要運作狀離。故於 圖所示之機率模_練資料1G中，檯燈之主要運作狀態為開啟，、電= 扇之主要運作狀態為中風量，電鍋之主要運作狀態為關閉。且各電器 之間使關性’㈣燈為例’於檯賴啟時，電風扇設為中風量 的機率高於電鋼設為保溫的機率’而f||組合轉換的順序性，於棱燈 與電風扇共同使用時，下個時間點為同樣檯燈與f風扇組合使用的機 201207654 及電、電風扇、電鋼等電器之間使用的相關性以 的特徵’係將計算出每—機率模型訓練資料 資料1〇0 同時參閱第二圖與三圖所示，將依據機率模型钏練 1〇之電器與^徵向量，計算相對應機率模型訓練資料 依特電錄11順序特徵方程式22及電器使用相 器你田ί式表不為機率模型訓練資料1〇之時間序列，以代表電 120 =1性及電&纟且合之順序性。以麵為例，其電轉徵方程 ^ 22後主不為(開啟）：1 :欠；（開啟）：3次。其電器狀態順序特徵方程 _啟顧）：3次。其電器使_鋪财程式24區 ίι閉合與油依電脉合，f 11㈣組合絲蝴開啟， 電風欠，（開啟相量，關閉）：3次。兩相依電器組合檯燈與 二 异出Ϊ—解_訓練資料1()之_序列，再祕由時間序列 i .代表每—機率模型訓練資料1G之電器使用狀態機率值，此電器 態機率值與時間序列係將滿足如下列式⑴所示： Ρλ (1)； 7exp ΣΆ + Σ七' 其中，ΣΆ、 、τ 7 表示電器特徵方程式20，s/為同一時間每一電器 C電器狀_訊與舰向f組合，i為自錄，〜職〜之特徵榍 ，’特徵權重為每—電器之運作狀態與相對應之特徵向量組洽 、目關性比重。，A表示電器狀態順序特，方程式22，力為同一霄 器之電器狀態資訊轉換組合，^為自然數，/為對應"y•之順序權重值 順序權重麵騎—電紅運作變化的順序佩重。丨W表示電器孩 :相依特徵方程式24，&為同—時間電^彼關之電器狀態資訊的麵 ° k為自然數’又；為對應~之相依權重值，相依權重值係為電器之 整體運作狀態及任兩電H之運作狀態的相依性比重。Z為正規化名 201207654The equation, - «state order characteristic equation and - electrogram _ according to the characteristic equation, to represent the machine to _ practice (four) of the coffee, and then calculate the probability of the probability of the machine model training data by the coffee phase, to Establishing a loop electrical appliance to make a probability model. & The above is the thief's vertical circuit power supply Laijia overall step under the slogan will be on the road, with a prismatic electric fan and an electric steel as an example, for the step 4 integrated electrical status information and feature vector establishment _ road For details, please refer to the second and third figures. In the above step S14, the electromechanical riding and the feature vector are integrated according to the time segment into a plurality of machine __ materials, please refer to the second machine of the invention, the machine training data is not intended, as shown in the figure Each probability model training data 彳 (4) domain lamp fan, electric steel electrical cancer information and corresponding feature vector (not shown), and table lamp, electric fan, electric age electrical in the probability model training data 1〇 Most of the states estimated in this paper will represent the main operational status of the training data in this probability model. Therefore, in the probability model shown in the figure, the main operating state of the desk lamp is ON, and the main operating state of the fan is the amount of stroke, and the main operating state of the electric pot is off. And the electrical ('four) lights between the electrical appliances as an example] in the Taiwan Laiqi, the probability of the electric fan set to a stroke is higher than the probability that the electric steel is set to heat insulation' and f||combination conversion sequence, in the edge lamp and When the electric fan is used together, the next time point is the combination of the same table lamp and the f-fan 201207054 and the correlation between the electric, electric fan, electric steel and other electrical appliances. The system will calculate the per-probability model training. Data 1〇0 At the same time, as shown in the second and third figures, the electrical and eigenvectors of the 〇 , , , , , , , , , , , , , , 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器 电器The time you are in the field is not the time series of the probability model training data, to represent the electrical 120 =1 sex and electricity & Taking the surface as an example, the main conversion equation ^ 22 is not (open): 1 : owed; (open): 3 times. Its electrical state sequence characteristic equation _ Kai Gu): 3 times. Its electrical appliance makes the _ shopping program 24 area ίι closed with the oil according to the electric pulse, f 11 (four) combined silk butterfly open, electric wind owed, (open phasor, close): 3 times. The two-phase electrical combination table lamp and the two different Ϊ-solution _ training data 1 () _ sequence, re-secret by time series i. Represents each probability model training data 1G electrical use state probability value, the electrical state probability value and The time series system will satisfy the following formula (1): Ρλ (1); 7exp ΣΆ + Σ7' where ΣΆ, τ 7 represent the electrical characteristic equation 20, s/ is the electrical property of each electrical appliance at the same time_ Ship to f combination, i is self-recording, ~ job ~ characteristics 榍, 'feature weight for each - electrical operation state and the corresponding feature vector group, the proportion of the target. , A indicates the order of the electrical state, Equation 22, the force is the electrical state information conversion combination of the same device, ^ is the natural number, / is the order of the corresponding "quote" weight order, the weight face riding - the order of the red operating changes Pay attention.丨W indicates the electrical child: the dependence characteristic equation 24, & is the same as the time-electricity of the electrical state information of the face ° k is the natural number 'again; for the corresponding ~ the weight value, the dependent weight value is the electrical The overall operational status and the proportion of the dependence of the operating status of the two electric H. Z is a regular name 201207654
母，用以將式⑴中的指數項數值正規化成為介於〇到】之間的數值， 以滿足機率特性。巧為所求得之電器使用狀_率值4經由上式⑴ 計算每-機率模型訓練雜1Q之電器使用狀態機率值時，可藉由牛頓 法、梯度上升法、信念傳遞鮮法或反覆分驗算法等最佳化演算法 調整特徵權重值、順序權重值、相依權重值，以使得電器使用狀態機 率值為最大化。並將所求得之每，率模型訓練龍1Q的特徵權重 值、順序權重值、相依權重值以及電器使用狀態機率值經儲存後以建 立成為可供職迴路電器制狀態之迴路f ^使用狀態機率模型。 以上為步驟S14整合電n狀態資訊與特徵向量建立丨迴路電器使 用狀態機率模型的制，底下麟應用此迴路電器賴狀態機率模型 辨識迴路電器使用狀態進行說明。 第四圖為本發明之迴路電ϋ使職態機率模型顧於辨識迴路用 電狀態之流賴’如_示，首先，如步驟S2Q，糊迴路電錶_ 出迴路上之迴路用電資訊。之後’如步驟S22，娜迴路用電資訊的 電力特徵。最後，如麵S24，透過轉比法(v_A|gQn搭 配迴路電器使用狀態機率模型針對各別電器進行單—電器推論，以將 電力特徵對應至迴路電器錢狀態機率模型，所求得之最大機率解即 為電器使肖狀ϋ。制鼓叉分難算法(丨tefat|Ve aassjficatiQn A_hm)搭配迴路電器使用狀態機率模型，針對相依使用電器進行交 又電器推論’ 4認最終電㈣使賊態，以辨識出迴路 器使用狀態。 1 € 經由實_朗可知本個_由量化„使狀餐性及相關 性建立迴路電ϋ使職態機率模伽做為迴路電力資訊模型。 演算法搭配此觀可㈣表上靖之家庭電紐職況細分 每一電器的運作狀態。 项出 以上所述之實施例僅係為說明本發明之技術思想及特點，宜 在使熟習此項技藝之人士能夠瞭解本發明之内容並據以實施，^ 以之限林㈣之專_圍，即大歧本發日麟揭私精神所作之^ 201207654 ^ 紅飾’仍應涵蓋在本發明之專利範圍内。 【圖式簡單說明】 =一圓為本發明之建立迴路電力資訊模型之流程圖。 第=圖為本發明之機率模型訓練資料之示意圖。 第三圖為本發明之建立時間序列計算電器使用狀態機率值之示意圖。 第四圖為本發明之迴路電器使用狀態機率模型應用於辨識迴路用電狀 態之流程圖。 【主要元件符號說明】 10機率模型訓練資料 2〇電器特徵方程式 22電器狀態順序特徵方程式 24電器使用相依特徵方程式The mother is used to normalize the value of the exponential term in equation (1) to a value between 〇 to 】 to satisfy the probability characteristic. The value of the electric appliance used by the _ rate value 4 is calculated by the above formula (1). When the probability value of the electric appliance usage state of the per-probability model training miscellaneous 1Q is calculated, the Newton method, the gradient ascending method, the belief transfer method or the reverse division can be used. The optimization algorithm such as the algorithm adjusts the feature weight value, the sequence weight value, and the dependent weight value to maximize the probability of using the state of the appliance. And each of the obtained rate model training dragon 1Q feature weight value, order weight value, dependent weight value and electrical use state probability value is stored to establish a loop of the state of the available circuit electrical system f ^ use state probability model. The above is the system for establishing the state probability model of the loop circuit electrical appliance by integrating the electrical n state information and the feature vector in step S14, and the bottom lining uses the loop electrical state probability model to identify the use state of the loop electrical appliance. The fourth figure is the loop power of the present invention, and the activity probability model of the circuit is determined by the identification of the power state of the circuit. As shown in the figure, first, as in step S2Q, the circuit information of the circuit on the circuit of the paste circuit is output. After that, as in step S22, the electric characteristics of the electricity information are used. Finally, as shown in Fig. S24, through the conversion ratio method (v_A|gQn with the loop electrical appliance using the state probability model for individual electrical appliances to make a single-electrical inference to match the power characteristics to the loop electrical money state probability model, the maximum probability obtained The solution is the electric appliance to make the sound of the sound. The drum fork splitting algorithm (丨tefat|Ve aassjficatiQn A_hm) is matched with the state probability model of the loop electrical appliance, and the electric appliance is used for the dependent electrical appliance. 4 The final electric power (4) makes the thief state. In order to identify the state of use of the circuit breaker. 1 € According to the actual _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ (4) The operating status of each appliance is subdivided by the family's home appliance status. The above-mentioned embodiments are only for explaining the technical idea and characteristics of the present invention, and it is desirable to enable those skilled in the art to understand the present invention. The content of the invention is implemented according to the fact that it is limited to Lin (4), which is the speciality of the Daqi Benfa. [Simplified illustration] = one circle is the flow chart of the loop power information model of the invention. The figure is a schematic diagram of the probability model training data of the present invention. The third figure is the establishment of the time series calculation appliance of the present invention. Schematic diagram of the use of the state probability value. The fourth figure is a flow chart of the use of the circuit state probability model of the invention in the identification circuit power state. [Main component symbol description] 10 probability model training data 2〇Electrical characteristic equation 22 electrical state Sequential characteristic equation 24 electrical appliances use dependent characteristic equations