JP2011017674A - System and program for estimation of electrical apparatus operation state - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a system for estimation of an electrical apparatus operation state, capable of efficient estimation not requiring learning.SOLUTION: The electrical apparatus operation state estimation system is provided for estimating operation states of a plurality of electrical apparatus connected to the same feeder. The system includes a characteristic quantity extraction means for extracting information of the waveform of a current flowing through the feeder, and an individual apparatus information storage means for storing information of the current waveform when only an electrical apparatus is operated relative to each electrical apparatus. The system also includes an electrical apparatus operation state estimation means for acquiring estimation results of the operation states of the plurality of electrical apparatus by executing genetic algorithm wherein a combination of the operation states of the plurality of electrical apparatus is assumes as an individual gene arrangement, and one or a plurality of information of the current waveform are acquired from the individual apparatus information storage means based on the individual gene arrangement, and evaluation is performed by forming information to be compared with extracted information of the waveform of the current flowing through the feeder.

Description

   The present invention relates to an electrical equipment operating status estimation system and program, and for example, information on sensors attached to a common power supply line for a plurality of electrical equipment and electronic equipment (hereinafter, the electrical equipment and electronic equipment are collectively referred to as electrical equipment). Therefore, the present invention can be applied when estimating the operating status of a plurality of electrical devices.

  In an era when ecology is screaming like nowadays, it is important to understand how each electrical device is operated in a home or office. However, it is not practical to provide a sensor or the like for detecting the operation status for each electric device because the cost burden is great.

  For this reason, the system described in Patent Document 1 that estimates the operating status of a plurality of electrical devices using a single sensor has already been used as a method for estimating the operating status of a plurality of electrical devices that have been used by power consumers. Proposed. In this system, a measurement sensor that measures power supply current, power supply voltage, etc. is installed in the power supply trunk, and the operating status of each electrical device is determined using feature quantities such as harmonic components according to the operating status of each electrical device. presume. More specifically, in this system, the operating status of each electrical device is estimated by pattern recognition using a neural network based on the harmonics and phase differences of the power supply voltage and power supply current obtained from the measurement sensor. Yes.

JP 2000-292465 A

  However, in the system described in Patent Document 1, with respect to combinations of various operating states when each electric device operates independently, the data on the harmonic current measured in advance and the phase with respect to those voltages and the answer at that time It is necessary to learn the neural network in advance using the operating state of a certain electrical device as teacher data. Furthermore, it is necessary to relearn this neural network every time a new electric device is added. For this reason, there is a problem that the processing time required for constructing the neural network is enormous.

  The present invention has been made in view of the problems as described above, and an object of the present invention is to provide an electrical apparatus operating state estimation system and program that can be efficiently estimated with little or no learning. It is to provide.

  A first aspect of the present invention is an electrical equipment operation state estimation system for estimating an operation state of a plurality of electrical devices connected to the same power supply line. (1) A feature amount for extracting information on a current waveform flowing through the power supply line. Extraction means; (2) individual device information storage means for storing current waveform information when only the electrical device is operating for each of the electrical devices; and (3) operating status of a plurality of electrical devices. The current flowing through the feeder line extracted by the feature amount extraction unit by obtaining information on one or more current waveforms from the individual device information storage unit based on the individual gene sequence It is characterized by comprising an electrical equipment operation state estimation means for executing a genetic algorithm for forming and evaluating information to be compared with waveform information and obtaining an estimation result of operation states of a plurality of electrical equipments.

  The second aspect of the present invention is an electrical equipment operating status estimation program for estimating operating statuses of a plurality of electrical equipment connected to the same power supply line, and (1) information on a current waveform flowing in the power supply line. (2) individual device information storage means for storing current waveform information when only the electric device is operating for each of the electric devices, and (3) a plurality of pieces The combination of the operating statuses of the electrical equipment is an individual gene sequence, and based on the individual gene sequence, information on one or more current waveforms is obtained from the individual equipment information storage means, and the feature amount extraction means extracts the above It functions as an electrical equipment operation status estimation means that executes a genetic algorithm that forms and evaluates information to be compared with information on the current waveform flowing through the feeder, and obtains the estimation results of the operating status of multiple electrical equipment. And wherein the Rukoto.

  According to the present invention, it is possible to provide an electrical equipment operation state estimation system and program that can be efficiently estimated with little or no learning.

It is a block diagram which shows the functional structure of the electric equipment operating condition estimation system which concerns on 1st Embodiment. It is a current waveform diagram which shows the current waveform for every operation mode of a microwave oven. It is a current waveform diagram which shows the individual current waveform for every electric equipment, and the total current waveform which flows through a feeder. It is explanatory drawing of the gene arrangement | sequence in 1st Embodiment. It is a flowchart which shows the outline | summary of the process of the genetic algorithm which the electric equipment operating condition estimation part in 1st Embodiment is utilizing. It is explanatory drawing of the production | generation process of the initial group of the individual | organism | solid in 1st Embodiment. It is explanatory drawing of the selection process (tournament selection process) in 1st Embodiment. It is explanatory drawing of the 1 point crossover process in 1st Embodiment. It is a current waveform diagram which shows the current waveform of the fluorescent lamp and television which are similar. It is explanatory drawing of the mutation process in 1st Embodiment. It is a block diagram which shows the functional structure of the electric equipment operating condition estimation system which concerns on 2nd Embodiment. The frequency characteristic of the current waveform about a television is shown.

(A) 1st Embodiment Hereinafter, 1st Embodiment of the electric equipment operating condition estimation system and program by this invention is described, referring drawings. In the following description, it is assumed that the electrical equipment operation status estimation system according to the first embodiment is provided in a home, but it is needless to say that the installation target of the electrical equipment operation status estimation system is not limited to a home. .

(A-1) Configuration of First Embodiment FIG. 1 is a block diagram showing a functional configuration of an electrical equipment operating status estimation system according to the first embodiment.

  In FIG. 1, the electric equipment operating state estimation system 1 according to the first embodiment flows with a plurality of (N in FIG. 1) electric appliances 2-1,. It is inserted at the position of the feeder 3.

  The position of the power supply line 3 in which the electrical equipment operation state estimation system 1 is inserted may be, for example, the position of the power supply line with respect to the entire house, and the house is divided into a plurality of sections such as the first floor and the second floor. It may be the position of the power supply line for any of the sections when viewed as being configured. In addition, for example, the electrical equipment operation status estimation system 1 may be installed in the vicinity of the power supply switch 4 such as a breaker system, or the electrical equipment operation status estimation system 1 is configured by being integrated with the breaker system. There may be.

  The electrical equipment 2-n (n is 1 to N) is not only one that can be fed from the feeder 3 by plugging the equipment into an outlet, but also one that is installed in the house from the beginning good. Examples thereof include a vacuum cleaner, a microwave oven, a hair dryer, an iron, a toaster, a television (television receiver), a refrigerator, a fluorescent lamp, and an air conditioner. However, the type of the electrical equipment 2-n needs to have a reference waveform pattern registered in an electrical equipment feature quantity database (electric equipment feature quantity DB) 14 described later. In the case of the first embodiment, for example, when there are two televisions, each television is a different electric device.

  Functionally, the electrical equipment operation status estimation system 1 functionally includes a measurement sensor 11, a feature quantity extraction unit 12, an electrical equipment operation status estimation section 13, an electrical equipment feature quantity DB 14, and an estimation result output section 15. Here, the feature quantity extraction unit 12 includes an analog / digital conversion unit (AD conversion unit) 21 and a waveform information extraction unit 22.

  All or part of the waveform information extraction unit 22, the electrical equipment operating state estimation unit 13, the electrical equipment feature value DB 14, and the estimation result output unit 15 may be realized by a program and a computer that executes the program. .

  The measurement sensor 11 measures the current and voltage at the position of the feeder 3 where the operating current of the electrical devices 2-1,... This is output to the quantity extraction unit 12.

  The feature amount extraction unit 12 extracts a feature amount (superposed waveform of an operating current waveform described later) corresponding to a combination of electric devices that are in an operating state at that time.

  The AD converter 21 in the feature quantity extractor 12 converts the current waveform signal and the voltage waveform signal, which are analog signals from the measurement sensor 11, into digital signals. Here, the sampling rate is arbitrary, but it may be selected in consideration of an operating current having a high frequency component having the highest meaning among individual operating currents for each electric device.

  The waveform information extraction unit 22 starts the current waveform synchronized with the next 0 cross point of the voltage waveform signal (digital signal) from the time of the current waveform signal (digital signal) synchronized with the 0 cross point of the voltage waveform signal (digital signal). The current waveform pattern up to the time of the signal (digital signal) is extracted as a feature amount corresponding to the combination of electric devices that are in operation at that time.

  Some of the electrical devices 2-1 to 2-N include capacitive loads and inductive loads. Therefore, the phase of the weight current (superimposed operating current) of the operating current determined by the combination of the operating electrical devices is , Not constant with respect to the phase of the supply voltage, but changes with respect to the combination of electrical devices in operation. Moreover, the operating current waveform for each of all the electrical devices 2-1 to 2-N has the property that the waveform of the power supply voltage frequency is contained. Therefore, rather than cutting out the waveform from the change in the superimposed operating current itself, the cutout period from the superimposed operating current is determined and cut out based on the voltage waveform of the stable sine wave waveform of 50 Hz or 60 Hz, and the cutting accuracy is improved. I am trying.

  Here, the waveform information extraction unit 22 obtains a plurality of current waveform patterns having different periods (periods between 0 crosses before and after the voltage waveform), and synthesizes them by aligning the time axes of the plurality of current waveform patterns. The noise may be reduced by obtaining an average, and the current waveform pattern after the noise reduction may be extracted as a feature amount corresponding to the combination of electric devices that are in operation at that time.

  In addition, the waveform information extraction unit 22 confirms that a plurality of current waveform patterns to be used for synthesis or averaging have a predetermined similarity (for example, the sum of squared differences is within a threshold value), and is used for synthesis or averaging. In addition, even when the synthesis process and the averaging process are not executed, the current waveform pattern to be extracted as the feature amount and the current waveform pattern in the immediately preceding period and the immediately following period have a predetermined similarity. And the extraction may be validated. In many electrical devices, when changing from on to off or from off to on, a current having a waveform different from that in the normal state flows due to a transient phenomenon. It is preferable that the waveform information extraction unit 22 obtains a feature amount (current waveform pattern) so as to eliminate the influence of such an excessive period.

  The electrical device feature DB 14 stores individual current waveform patterns for the electrical devices 2-1,..., 2-N (see FIG. 3 (A1) to FIG. 3 (A3) described later). The individual current waveform pattern is a pattern (digital signal) of a portion cut out from the current waveform obtained when only the electric device 2-n is operating based on the zero cross of the voltage waveform. The individual current waveform pattern stored in the electric device feature value DB 14 may be provided by the manufacturer of the electric device 2-n, a consumer group, or the like. The current waveform pattern output from the waveform information extraction unit 22 may be used when the learning mode is set and only the electric device 2-n to be learned is in an operating state (other electric devices are in a non-operating state). . If the waveform information extraction unit 22 is characterized by combining or averaging current waveform patterns of a plurality of periods, the individual current waveform pattern stored in the electrical equipment feature value DB 14 is also in accordance with such processing. Shall.

  Some electric appliances have greatly different current waveforms depending on operation modes such as strength and weakness. For example, as shown in FIG. 2, the microwave oven has a greatly different current waveform amplitude between the “1500 W” mode and the “500 W” mode. In the case of the first embodiment, for an electric device whose current waveform varies greatly depending on the operation mode, an individual current waveform pattern for each operation mode is stored in the electric device feature amount DB 14.

  When there are exactly the same products (products of the same manufacturer and the same serial number) in the electrical devices 2-1 to 2-N, the same individual current waveform patterns as many as the number existing in the electrical device feature DB 14 exist. In addition, one individual current waveform pattern is stored in the electrical equipment feature value DB 14, and the individual current waveform pattern is repeatedly read out according to the number of operating states. Also good.

  The electrical equipment feature value DB 14 is intended to be provided for each electrical equipment operating state estimation system 1. However, an external server or the like may have the electric device feature value DB 14, and the plurality of electric device operation state estimation systems 1 may share the electric device feature value DB 14.

  When the current waveform pattern (feature amount) extracted by the waveform information extraction unit 22 combines a plurality of individual current waveform patterns stored in the electric device feature amount DB 14, the electrical equipment operating state estimation unit 13 By searching for the most similar, the combination of electric devices in operation is estimated.

  In the case of the first embodiment, the electrical equipment operating state estimation unit 13 is characterized by using a genetic algorithm (GA) for estimation processing. The details of the estimation process (genetic algorithm) in the electrical equipment operation status estimation unit 13 will be clarified in the “Description of Operation” section.

  The estimation result output unit 15 outputs a combination (estimation result) of the electric devices in the operating state obtained by the electric device operating state estimation unit 13. The output method may be any of display output, print output, storage output, transmission output to another device (for example, a higher-level monitoring device), and a plurality of output methods may be applied.

  For example, display elements such as LEDs corresponding to each electrical device (each operation mode for a plurality of operation modes) may be prepared, and only the display elements of the electrical devices in the operating state may be lit.

  In addition, for example, when the estimation operation of the electrical equipment operation status estimation system 1 is periodically performed at predetermined time (for example, 1 minute) intervals, the estimation result at each time may be stored as it is. Instead, the start time of the on state and the end time of the on state may be recorded for each electric device (for each operation mode when there are a plurality of operation modes).

  In addition, it may be transferred to a personal computer terminal connected by wire or wirelessly, a home appliance such as a television, a remote controller, a mobile phone, or the like so that display or storage is executed on the transfer destination device. Further, it may be sent to a control device such as a power control device. For example, it may be sent to an air conditioner and the air volume may be changed from “strong” to “weak” by collating the estimation result with a predetermined condition. Furthermore, it may be transmitted to an electric power company, various monitoring servers, etc. via a network.

(A-2) Operation of the First Embodiment Next, the operation of the electrical equipment operating status estimation system according to the first embodiment will be described with reference to the drawings.

  The electrical equipment operation status estimation system 1 starts an estimation operation when an operator operates an estimation activation key (not shown). In addition, when the estimation operation is automatically started at a predetermined cycle (for example, every minute), the electrical equipment operation state estimation system 1 uses a built-in timer (not shown) for a new estimation operation. The estimated operation starts at the start time of.

  When the electrical equipment operation status estimation system 1 starts a new estimation operation, the measurement sensor 11 measures the current and voltage at the position of the feeder line 3 through which the currents of all the electrical equipments in operation are flowing. It is output to the feature quantity extraction unit 12. The AD converter 21 of the feature quantity extraction unit 12 converts the current waveform and voltage waveform, which are analog signals from the measurement sensor 11, into digital signals, respectively, and then the waveform information extraction unit 22 of the feature quantity extraction unit 12 A current waveform pattern representing a current waveform in a period of one cycle of the voltage waveform synchronized with the measured voltage waveform is extracted as a feature amount and provided to the electric equipment operating state estimation unit 13.

  When the extracted current waveform pattern (feature amount) combines a plurality of individual current waveform patterns stored in the electric device feature amount DB 14 in the electrical device operating state estimation unit 13 of the first embodiment. Is searched using a genetic algorithm to estimate the combination of electrical devices in operation.

  That is, the electrical equipment operating state estimation unit 13 estimates which electrical equipment 2-n is used and in an operating state by measuring the total current flowing through the feeder line 3. At this time, the total current to be measured is the sum of the currents flowing through the electrical devices in the on state (operating state), and has information obtained by combining the operating states of the electrical devices. In order to grasp the operating state of each electrical device from the total current information, it is necessary to solve the inverse problem of estimating the input (the operating state of each electrical device) from the output result (total current information).

  FIGS. 3A1 to 3A3 show individual current waveform patterns for the vacuum cleaner (VC), the hair dryer (HD), and the television (TV), respectively. If only these three types of electric devices (VC, HD, TV) are in operation, the current waveform pattern for the total current flowing through the feeder 3 is the individual current shown in FIGS. 3 (A1) to (A3). A current waveform pattern as shown in FIG. The input to the electrical equipment operation status estimation unit 13 is a current waveform pattern as shown in FIG. 3B, and the electrical equipment operation status estimation unit 13 is obtained from the current waveform pattern as shown in FIG. The individual current waveform pattern that is an element of the current waveform pattern, in other words, an electric device having the individual current waveform pattern is defined.

  The current waveform of an electric device has various properties (a) to (e) as described below, and has unique characteristics for each operating electric device.

(A) Waveforms are similar depending on components incorporated in electrical equipment (b) Amplitude (current value) varies greatly depending on electrical equipment (c) Waveform changes greatly depending on the operation mode such as strength (d) Power supply frequency Contained in the waveform of electrical equipment (e) Odd-order harmonic components of the power supply frequency are contained by the electrical equipment. Therefore, it is not easy to determine the individual current waveform pattern that is the element from the current waveform pattern of the total current, Apply an algorithm to find the optimal solution. The number (N) of electric devices to be combined is quite large (for example, the number of electric devices in one home is about 30 to 40), learning processing is unnecessary or simple, and database storage In view of the ability to suppress the capacity, in the first embodiment, the genetic algorithm is applied to determine the individual current waveform pattern as an element from the current waveform pattern of the total current. .

For example, the number of on / off combinations of N electrical devices, in other words, the number of types of total current is ^2N . When the reference waveform for the total current is stored in the database and collated, 2 ^N types of reference waveforms must be stored, and the number of types of reference waveforms is so large that it cannot actually be stored. As described above, in the case of the first embodiment to which the genetic algorithm is applied, the number of individual current waveform patterns stored in the electrical device feature DB 14 is N, which is the number of electrical devices. In the case of an electric device whose current waveform varies greatly depending on the operation mode such as strength and weakness, the individual current waveform pattern for each operation mode is stored in the electric device feature value DB 14, but this is also necessary for the electric device feature value DB 14. The storage capacity is practically not a problem.

  FIG. 4 is an explanatory diagram of a gene sequence that identifies individuals necessary for the genetic algorithm. In the case of the first embodiment, the gene sequence is a bit string (hereinafter referred to as a gene bit string) that takes “0” (represents a non-operating state) or “1” (represents an operational state). For an electric device that does not have an operation mode or an electric device that has an operation mode but the current waveform hardly changes in the operation mode, one bit in the gene bit string is assigned. For electrical devices whose current waveforms vary greatly depending on the operation mode, the number of bits equal to the number of operation modes is assigned in the gene bit string. In the case of such an electric device, whether or not it is in an operating state is detected, but whether it is a non-operating state or an operating state in which operation mode is detected.

  In the case of the first embodiment, the gene bit string is, for example, an electric device having a larger current waveform amplitude (or an operation mode of a certain electric device) as it goes to the left, and the current waveform amplitude as it goes to the right. Is a small electric device (or an operation mode of an electric device). This is because the arrangement according to the amplitude of the current waveform increases the ease of the crossover process described later. For example, the vacuum cleaner (VC), the hair dryer (HD), and the television (TV) shown in FIGS. 3A1 to 3A3 have amplitudes of 35 to −35 [A], 18 to −18 [A], Since it is 4 to -4 [A], among these three types, the bit position for the vacuum cleaner is located to the left of the bit position for the hair dryer, and the bit position for the hair dryer is from the bit position for the television. Located on the left side.

  FIG. 5 is a flowchart showing an outline of the process of the genetic algorithm executed by the electrical equipment operation status estimation unit 13. The genetic algorithm is an algorithm that mimics the mechanism of biological evolution, and includes genetic operations such as selection / selection, crossover, and mutation.

  Also in the genetic algorithm executed by the electrical equipment operation status estimation unit 13, first, a plurality of individuals as solution candidates are generated as an initial population (step S101), and the input (current of the total current) is input to the plurality of individuals. It is evaluated whether there is an individual that can be evaluated as conforming to (waveform pattern) (step S102). When such an individual is obtained, the genetic algorithm is terminated. If such an individual cannot be obtained, genetic operations such as selection (step S103), crossover (step S104), and mutation (step S105) are repeated until such an individual is obtained. And search for the individual with the optimum solution. Selection, crossover, and mutation are processing for generating a new individual population.

  The processing of each step in FIG. 5 will be described in detail. In the following description of the processing steps, for simplicity of explanation, the gene bit string is 8 bits, and each bit position from the left side is a vacuum cleaner (VC), a microwave oven (MO), a hair dryer (HD), an iron, respectively. (IR), Toaster (TO), Television (TV), Refrigerator (RF), Fluorescent lamp (FL) operation / non-operational state (the number of electrical devices is less than the actual home, the operation mode Not to reflect the difference).

Initial population generation (step S101)
A set of a predetermined number of individuals is generated as an initial group. Individuals belonging to the initial population are determined by generating random numbers, for example. FIG. 6 is an explanatory diagram of generation of an initial group. Since the above-described gene bit string is applied as the gene sequence, an initial population may be formed using a binary random number generator. However, since there are few binary random number generators, as shown in FIG. 6, a decimal random number generator is used to generate a decimal random number and then convert the binary number into a binary number. Random numbers (individual gene bit strings) may be obtained.

  FIG. 6 shows a case where the number of individuals belonging to the initial group is 5, but it goes without saying that the number of individuals belonging to the initial group is not limited to this. For example, “individual 1” in FIG. 6 includes a vacuum cleaner (VC), a hair dryer (HD), an iron (IR), a television (TV), a refrigerator (RF), a microwave oven (MO), a toaster ( TO) and fluorescent lamp (FL) indicate a non-operating state.

  Although FIG. 6 shows that all the individuals belonging to the initial population are generated using random numbers, all or some of the individuals may be generated without using random numbers. When the electrical equipment operation status estimation system 1 according to the first embodiment automatically starts the estimation operation at a predetermined cycle (for example, every minute), the individual of the immediately preceding estimation result, the immediately preceding Individuals that have been subjected to genetic operations such as crossover and mutation may be included in the individuals belonging to the initial population.

Individual evaluation (step S102)
It is evaluated whether each individual belonging to the current processing target group conforms to the current waveform pattern extracted by the waveform information extraction unit 22 (hereinafter referred to as a characteristic current waveform pattern).

The individual current waveform patterns of all electric devices (or operation modes) whose value is “1” in the evaluation target individual (bit string) are extracted from the electric device feature amount DB 14 and synthesized (see FIG. 3). A combination of individual current waveform patterns is referred to as a combined current waveform pattern. Then, the fitness is calculated from the characteristic current waveform pattern extracted by the waveform information extraction unit 22 and the combined current waveform pattern. The degree of fitness may be arbitrarily determined, but for example, the following calculation formula can be applied.

  As shown in the equation (1), the time average value of the absolute difference between the characteristic current waveform pattern fe (t) and the combined current waveform pattern fd (t) at each time t is set as a value x representing similarity, and this value The fitness f is obtained by applying the normalization function shown in Equation (2) to x. The degree of fit f obtained in this way is a value between 0 and 1, with 1 representing the complete match between the characteristic current waveform pattern and the combined current waveform pattern.

  Instead of the time average value of the absolute difference shown in the equation (1), the square root of the time average value of the square of the difference may be used as a value representing the similarity. When the optimization function is applied, there is a tendency that it is easy to take a value close to 1 even if the similarity is not so high as compared with the above-described fitness f, and in order to improve the evaluation accuracy, the expressions (1) and (2) The degree of fitness f to which is applied is used. In the equation (2), μ is a fixed parameter.

  It is determined whether there is a degree of fitness for each individual that exceeds a threshold that is set to a value that can be regarded as a match between the characteristic current waveform pattern and the combined current waveform pattern. If there is even one individual whose fitness exceeds the threshold, the individual with the highest fitness is used as the estimation result. It is estimated that the electrical device (or the operation mode of the electrical device) corresponding to the bit position to which “1” is assigned in the individual bit string of the estimation result is in the operating state.

  If the fitness for all individuals does not exceed the threshold value, the process proceeds to selection processing (step S103).

  Even if the number of executions of step S102 reaches the preset upper limit number, even when an individual whose fitness level exceeds the threshold cannot be found, the series of processes shown in FIG. 5 may be terminated. good. In this case, the estimation result may indicate that the estimation cannot be performed, or the individual having the maximum fitness obtained in the process of the last step S102 may be the estimation result. You may make it make the estimation result the individual which concerns on the largest thing in the maximum fitness obtained by the process of step S102.

Selection (step S103)
Selection (also referred to as regeneration) is an operation that models the natural selection of living organisms, and is an operation that leaves individuals who can be judged to be relatively good in the population based on the fitness to the next generation. As selection methods, there are roulette selection, expected value selection, ranking selection, tournament selection, elite preservation selection, and the like, and any selection method may be applied. Furthermore, if there is a possibility that the individual solution concentrates on a specific solution as the generation progresses, it is possible to avoid this problem by using a method such as sharing. It is. Below, the case where a tournament selection method is utilized is demonstrated.

  FIG. 7 is an explanatory diagram of the tournament selection process. The tournament selection method is a method in which a combination of individuals is randomly generated from the individual population so far, and an individual having a high fitness is selected from the combination. The predetermined number of selection (tournament size) is 2. In this case, individuals are randomly extracted from the individual population shown in FIG. 7 (A), and two tournaments as shown in FIG. 7 (B) are created, and the individual having the highest fitness for each tournament is selected. To do.

Crossover (step S104)
Crossover is an operation of generating two individuals that may have higher fitness than the individual before the two individuals are taken out of the individuals selected by selection and multiplied. The crossover method includes one-point crossover, multipoint crossover, uniform crossover, and any crossover method may be applied. Below, the case where 1 point crossing and multipoint crossing are utilized is demonstrated.

  With one-point crossover, one crosspoint is defined (one point), the first half of the point in the original individual (parent individual) is copied as is, the latter half is replaced with the second half of the other parent individual, and a new one This is an operation for generating a simple individual (child individual). FIG. 8 is an explanatory diagram of one-point crossover.

  In the example shown in FIG. 8, the crossover point is selected between the lower 3 bits and the lower 2 bits, and the lower 2 bits of the parent individual PA and the lower 2 bits of the parent individual PB are exchanged. Two child individuals CA and CB are generated. The individual population includes not only parent individuals but also child individuals, and the number of individuals increases. Here, the crossing point may be determined in advance, or may be determined each time by a random number or the like.

  As described above, the gene bit string is an electrical device (or an operation mode of a certain electrical device) having a smaller current waveform amplitude as it goes to the right. When the processing loop consisting of steps S102 to S105 is repeated several times, the current waveform having a high contribution ratio to the total current (in other words, the characteristic current waveform pattern extracted by the waveform information extracting unit 22) flowing through the feeder 3 is high. The bit string portion of an electric device having a large amplitude is generally correct, and there is a high possibility that the bit string portion having a small current waveform amplitude is erroneous. Generating an individual in which the lower bit string part is changed while using one-point crossover can generate a large number of individuals that differ only in the bit string part having a small current waveform amplitude, and can be expected to speed up the search for the optimum individual.

  Multipoint crossing means that multiple crossover points (multipoint) are determined, and the odd-numbered point and the next even-numbered point (if there is no next even-numbered point) in the original individual (parent individual) Is an operation of generating a new individual (child individual) by replacing only the space between the end position of the bit string and the same part of another parent individual. The crossing points in the multipoint crossing may be determined in advance, or may be determined each time by a random number or the like.

  In view of the nature of the current waveform of the electrical device, it is preferable to apply a predetermined crossover point in multipoint crossover. As one property of the current waveform of an electric device, there is a property that the waveform is similar depending on components (for example, a discharge tube and a heat source) incorporated in the electric device. 9A shows the current waveform of the fluorescent lamp, and FIG. 9B shows the current waveform of the television. Although not shown, the current waveforms of the toaster (first mode), the dryer (first mode), and the iron are similar. If the processing loop consisting of steps S102 to S105 is repeated several times, it approaches the correct answer, but although the amplitude is different, there is a possibility that an electric device having a similar waveform is obstructing the search. For this reason, it is preferable to set a crossover point in advance and generate a new individual by multipoint crossover so that both the bit positions of electrical devices having similar current waveforms are included.

Mutation (Step S105)
Mutation is an operation that randomly selects a bit string of a certain individual and changes the information. The purpose of performing the mutation is to prevent the individual from being settled to the local optimum solution and to search for the optimum solution in a wider range. FIG. 10 is an explanatory diagram of mutation. For all individuals generated by the processing up to the crossover or for each of a predetermined number of individuals determined using random numbers, the numbers determined by random numbers and the values of bit positions are changed to opposite values. In FIG. 10, the bit values of the third and seventh bits of the original individual PX are inverted to generate a mutated individual PY. Individuals PX and PY are both elements of the individual group.

  When a new individual population is generated by selection, crossover, and mutation, the evaluation in step S102 is executed again.

  As described above, for electrical devices whose current waveforms vary greatly depending on the operation mode, a bit for each operation mode is provided in the gene bit string. Such a plurality of bits for each operation mode cannot take “1” (operating state) at the same time. In an individual newly generated by crossover or mutation, the handling when a plurality of bits for each operation mode are simultaneously “1” is determined in advance, and the newly generated individual is handled accordingly. For example, such an individual may not be a group element. Further, for example, “1” is maintained for the operation mode determined by applying a random number or the like, “1” of the other operation mode is forcibly converted to “0”, and the individuals after such operation are grouped. It may be included.

  When the estimation result (combination of electric devices in the operating state) is obtained as described above, the estimation result output unit 15 outputs the estimation result.

(A-3) Effect of the First Embodiment According to the first embodiment, the waveform of the observed total current flowing through the feeder line, and the current waveform specific to the electrical device registered in the electrical device feature DB Since a genetic algorithm is applied to estimate an electric device in an operating state, parameter learning as in the case of using a neural network for estimation can be made unnecessary.

  In addition, when a target electric device is added, the current waveform of the additional electric device is added to the electric device feature amount database, and the bit length of the gene bit string is increased to change the allocation of each bit position. It can be easily handled. Incidentally, the method of using a neural network for estimation requires re-learning that requires a large amount of calculation at the time of addition, and cannot be easily added.

  Furthermore, according to the first embodiment, it is possible to make a portable system that can be reduced in size and can be freely connected to a power supply line or an electric device.

  Furthermore, by providing a download server that can be downloaded to the electrical device feature DB, and making the number of registered electrical devices public for each registrant, etc., power consumers and manufacturers can actively It is also possible to build an environment for registration.

(B) 2nd Embodiment Next, 2nd Embodiment of the electric equipment operating condition estimation system and program by this invention is described, referring drawings.

  In the first embodiment, the current waveform is used as the feature quantity of the electric device according to the operating state. However, in the second embodiment, the current value is used as the feature quantity of the electric device according to the operating state. The power (frequency characteristic) for each frequency component of the waveform is applied.

  FIG. 11 is a block diagram showing a functional configuration of the electrical equipment operating status estimation system according to the second embodiment. The same and corresponding parts as those in FIG. 1 according to the first embodiment are indicated by the same reference numerals. It is attached.

  In FIG. 11, the electrical equipment operating status estimation system 1A of the second embodiment includes a measurement sensor 11, a feature quantity extraction unit 12A, an electrical equipment operating status estimation unit 13, an electrical equipment feature quantity DB 14, and an estimation result output unit 15. The feature amount extraction unit 12A includes an analog / digital conversion unit (AD conversion unit) 21 and a frequency information extraction unit 23. Hereinafter, differences from the first embodiment will be described.

  The measurement sensor 11 according to the second embodiment measures a current flowing through the feeder line 3, and the analog / digital conversion unit 21 converts a current waveform composed of an analog signal into a digital signal.

  The frequency information extraction unit 23 performs Fourier transform on the current waveform signal converted into a digital signal, and calculates power for each frequency component. Such a power distribution for each frequency component is an observed feature amount.

  In the electrical device feature DB 14 of the second embodiment, the power (frequency characteristic) for each frequency component obtained by performing Fourier transform on the individual current waveform for each electrical device 2-1, ..., 2-N. Stored. FIG. 12 shows frequency characteristics for a television (TV).

  The electrical equipment operation status estimation unit 13 according to the second embodiment uses the frequency characteristics (features) extracted by the frequency information extraction unit 23 to obtain a plurality of individual frequency characteristics stored in the electrical equipment feature quantity DB 14. By searching for the most similar when combined, the combination of electric devices in operation is estimated. Although the feature amount is different from that in the first embodiment, the electric equipment operating state estimation unit 13 obtains an estimation result by the same genetic algorithm as in the first embodiment.

  The arrangement of the electrical devices assigned to the gene bit strings in the second embodiment may be determined according to the amplitude of the current waveform as in the first embodiment, and the maximum power among the power for each frequency component may be determined. It may be arranged according to the size.

  The estimation result output unit 15 is the same as that of the first embodiment.

  As described above, since the feature amount is different from that of the first embodiment, the feature amount extraction processing and the like are different. However, the entire operation is the same as that of the first embodiment, and thus the second embodiment. Description of the operation in is omitted.

  Also according to the second embodiment, the effect of applying the genetic algorithm is the same as that of the first embodiment.

  According to the second embodiment, since the power (frequency characteristic) for each frequency component of the current waveform is processed as a feature amount, it is not necessary to consider the phase component in the waveform cutting, and the feature amount extraction process Can be simplified.

  Further, according to the second embodiment, since the power (frequency characteristic) for each frequency component of the current waveform is used as the feature amount, it is possible to select the frequency component to be used for processing by the filter processing. It is also possible to improve SN by increasing the SN in the feature value by making a simple selection.

(C) Other Embodiments In the description of the above-described embodiment, various modified embodiments have been referred to. However, modified embodiments as exemplified below can be cited.

  In each of the above embodiments, the genetic manipulation for newly generating an individual population has been shown to be crossover and mutation, but other genetic manipulations may be used. For example, a new individual may be generated by exchanging values between two bit positions of a certain individual.

  In each of the above embodiments, the gene sequence is a bit string of “0” or “1” and the bit position for each operation mode is prepared. However, other gene sequences may be applied. May be. For example, the number of electric devices is applied as the number of elements of the gene sequence, and the value that can be taken by each element is set to a value corresponding to the number of operation modes of the electric device having the most operation modes. For example, if an electric device having three types of operation modes is an electric device having the largest number of operation modes, “0”, “1”, “2”, “3” are assumed to be values that can be taken as array elements, and the operation “0”, “1”, and “2” are treated as non-operating states and “3” are treated as operating states for one type of electric device, and “0” and “1” are used for two types of operating modes. "Is not in operation," 2 "is in operation in the first operation mode," 3 "is in operation in the second operation mode, and" 0 "is used for electrical devices with three operation modes. Non-operating state, “1” is treated as an operating state in the first operating mode, “2” is treated as an operating state in the second operating mode, and “3” is treated as an operating state in the third operating mode. The individual feature amount may be extracted from the feature amount DB 14.

  DESCRIPTION OF SYMBOLS 1, 1A ... Electrical equipment operation condition estimation system, 2-1 to 2-N ... Electrical equipment, 3 ... Feed line, 11 ... Measurement sensor, 12, 12A ... Feature-value extraction part, 13 ... Electrical equipment operation condition estimation part, DESCRIPTION OF SYMBOLS 14 ... Electric equipment feature-value database (electric equipment feature-value DB), 22 ... Waveform information extraction part, 23 ... Frequency information extraction part.

Claims (7)

In the electrical equipment operating status estimation system that estimates the operating status of a plurality of electrical equipment connected to the same feeder line,
Feature amount extraction means for extracting information on a current waveform flowing through the feeder line;
For each electrical device described above, individual device information storage means for storing current waveform information when only the electrical device is operating,
A combination of operating statuses of a plurality of electrical devices is used as an individual gene sequence, and based on the individual gene sequence, information on one or more current waveforms is obtained from the individual device information storage unit, and the feature amount extraction unit extracts And an electrical equipment operation status estimation means for executing a genetic algorithm for forming and evaluating information to be compared with the information on the current waveform flowing through the feeder line, and obtaining an estimation result of the operation status of a plurality of electrical equipment. Electrical equipment operating status estimation system characterized by   2. The electrical equipment operating state estimation system according to claim 1, wherein the current waveform information is a pattern of time change of the current waveform.   2. The electrical equipment operating status estimation system according to claim 1, wherein the current waveform information is a frequency characteristic of the current waveform.   The association of the electrical equipment with the elements in the gene sequence is arranged in accordance with the level of contribution of the current waveform of the electrical equipment to the current waveform flowing through the feeder line. The electrical equipment operation condition estimation system in any one of -3.   The electrical equipment operation status estimation system according to any one of claims 1 to 4, wherein the gene sequence is a bit string having different logical values in an operating state and a non-operating state.   For electrical devices having a plurality of operation modes having different current waveform information, a bit corresponding to each operation mode is prepared in the bit string constituting the gene sequence. The electrical equipment operation condition estimation system according to any one of claims 1 to 5. An electrical equipment operating status estimation program for estimating operating statuses of a plurality of electrical equipment connected to the same feeder line,
Computer
Feature amount extraction means for extracting information on a current waveform flowing through the feeder line;
For each electrical device described above, individual device information storage means for storing current waveform information when only the electrical device is operating,
A combination of operating states of a plurality of electrical devices is used as an individual gene sequence, and based on the individual gene sequence, information on one or more current waveforms is obtained from the individual device information storage unit, and the feature amount extraction unit extracts A genetic algorithm that forms and evaluates information to be compared with the information on the current waveform flowing through the power supply line, and functions as an electrical equipment operating status estimation means for obtaining an estimation result of operating statuses of a plurality of electrical equipment. A program for estimating the operating status of electrical equipment.

Images