Needs management system, power quality management system, and method

Abstract

To provide a power demand management system, a power quality management system, and a method for estimating power demand from information about consumers including a new power consumer, and utilizing the estimated power demand.SOLUTION: A demand management system includes consumer information complementing means for complementing information on an unknown item in consumer information including a plurality of items, existing consumer clustering means for extracting a consumer having information similar to that of an existing consumer in regard to information on existing consumers including a plurality of items, and classifying existing consumers, demand pattern generation means for generating demand patterns of new consumers from a demand pattern of a group of consumers having the similar information, and local demand amount calculation units that calculate, as estimated demand information, load amounts in respective local points of a power distribution system on the basis of the demand patterns of new consumers and the demand patterns of the existing consumers.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a demand management system, a power quality management system, and a method for estimating and utilizing power demand from information on customers including new power customers.

Background Art

[0002] In recent years, the environment surrounding the energy business has changed significantly, such as the increase in renewable energy, the medium- and long-term changes in power demand such as the decrease in power demand due to population decline, the change in the relationship between new business entities and existing power infrastructure operators, and the establishment of the power trading market. Power infrastructure operators are required to respond to these environmental changes. For this reason, due to the uneven distribution of customers in the distribution system, there are concerns about over-investment in distribution facilities, and the desire to suppress facility investment has been increasing. Therefore, in order to make the best use of the existing system and improve stability, environmental compatibility, and efficiency, it is necessary to convert to a new power network. Against this background, there is a need to appropriately grasp and guide the power demand of power customers. Furthermore, the need to provide new services to power customers thus grasped is also increasing.

[0003] As the background art in this technical field, there is Patent Document 1. Patent Document 1 discloses generating a plurality of power consumption models from consumption power information storing various information on the power consumption of each customer where a smart meter is installed, and estimating the power consumption amount by specifying a power consumption model that satisfies a similarity criterion from among the plurality of models for second consumption power information storing various information on each customer where a power meter is installed but a smart meter is not installed.

[0004] Further, Patent Document 2 discloses a distribution system state estimation system that estimates the state (active power, reactive power, voltage, etc.) of loads and power sources installed at nodes where measurement devices are not installed, using equipment information on the distribution line and measurement information by measurement devices at some nodes, and obtaining the PQ amounts at each point in the distribution system.

[0005] Furthermore, Patent Document 3 discloses that the actual power consumption of each electricity consumer is collected for a predetermined period, and for the load patterns of multiple consumers over the number of days in that period, characteristic parameter values that best represent the shapes of all the load patterns are calculated, and clusters are created based on the similarity of the characteristic parameter values, thereby obtaining representative load patterns from the long-term actual power consumption data accumulated in the past.

[0006] Furthermore, Patent Document 4 discloses that in a voltage monitoring control system, a voltage monitoring control device, a measuring device, and a voltage monitoring control method, without increasing the communication load, the voltage is maintained while following the voltage fluctuations in the distribution system, and an appropriate voltage range is commanded for transformer-type voltage control equipment, thereby estimating the load from the statistical values of smart meter data and determining the voltage range.

Prior Art Documents

Patent Documents

[0007]

Patent Document 1

Patent Document 2

Patent Document 3

Patent Document 4

Summary of the Invention

Problems to be Solved by the Invention

[0008] By using the method disclosed in Patent Document 1, it is possible to provide a power demand matching means including a customer extraction means for extracting customers similar to other existing customer information, an existing customer load pattern clustering section for classifying customers according to the load pattern of power consumption, a means for estimating the power consumption of customers, and a customer load pattern matching section for matching the estimated power consumption with customers having a similar load pattern among the set of customers extracted by the customer extraction means.

[0009] However, the method described in Patent Document 1 does not disclose having a PQ amount calculation section for calculating the PQ amount at each point of the distribution system based on the matching result, and a power demand sensitivity analysis section for correcting the load pattern of customers based on the calculated PQ amount and determining the distribution facility plan.

[0010] By using the method disclosed in Patent Document 2, it has a PQ amount calculation section for calculating the PQ amount at each point of the distribution system based on the matching result, and a power demand sensitivity analysis section for correcting the load pattern of customers based on the calculated PQ amount and determining the distribution facility plan. However, the method described in Patent Document 2 does not disclose means based on customer information.

[0011] By using the method disclosed in Patent Document 3, it is possible to provide power demand matching by a customer load pattern matching section that matches the estimated power consumption with customers having a similar load pattern among the set of customers extracted by the customer extraction means. However, the method described in Patent Document 3 does not disclose having a customer extraction means for extracting customers similar to other existing customer information, an existing customer load pattern clustering section for classifying customers according to the load pattern of power consumption, a means for estimating the power consumption of customers, a PQ amount calculation section for calculating the PQ amount at each point of the distribution system based on the matching result, and a power demand sensitivity analysis section for correcting the load pattern of customers based on the calculated PQ amount and determining the distribution facility plan.

[0012] By using the method disclosed in Patent Document 4, it is possible to maintain the voltage following the voltage fluctuations in the power distribution system without increasing the communication load while utilizing the smart meter data, and to command an appropriate voltage range for the transformer type voltage control device, enabling centralized voltage management.

[0013] However, the smart meter data of new customers is scarce and it is difficult to estimate the load, and a method for accurately estimating the load of new customers has not been disclosed.

[0014] In the aforementioned Patent Documents 1, 2, 3, and 4, means for simultaneously satisfying the needs of realizing, on the one hand, the characteristics of the power demand of customers with high accuracy under more uncertain conditions, including the specific extraction method of similar customers and the means for correcting the customer load pattern, and, on the other hand, the response to the information provision to customers, as well as the system, have not been disclosed.

[0015] The present invention has been made in view of the above problems, and an object thereof is to provide a demand management system, a power quality management system, and a method for estimating and utilizing power demand from information on customers including new power customers.

Means for Solving the Problems

[0016] From the above, in the present invention, "a demand management system characterized by having customer information complementing means for complementing information on unknown items for customer information composed of a plurality of items, existing customer clustering means for extracting customers having information similar to the information of existing customers and classifying existing customers for customer information composed of a plurality of items, demand pattern generation means for generating a demand pattern of a new customer from the demand patterns in a set of customers having similar information, and a demand amount calculation unit for each location that calculates the load amount at each location in the power distribution system as demand assumption information based on the demand pattern of the new customer and the demand pattern of the existing customer".

[0017] Also, in the present invention, there is provided "a power quality management system that manages the power quality of a distribution system using demand assumption information, which is the load amount at each point of the distribution system obtained in a demand management system, wherein the power quality management system includes a control device for an SVC that is arranged in the distribution system and adjusts the secondary side tap position according to a setting value, and the demand assumption information, which is the load amount at each point of the distribution system, is used for calculating the setting value."

[0018] Also, in the present invention, there is provided "a power quality management system that performs power demand prediction for a distribution system using demand assumption information, which is the load amount at each point of the distribution system obtained in a demand management system, wherein the power quality management system uses the demand assumption information, which is the load amount at each point of the distribution system, when performing power demand prediction for the distribution system."

[0019] Also, in the present invention, there is provided "a demand management method characterized by complementing information on unknown items for information on consumers composed of a plurality of items, extracting consumers having information similar to that of existing consumers for information on existing consumers composed of a plurality of items, classifying existing consumers, generating a demand pattern for new consumers from the demand patterns in the set of consumers having similar information, and calculating the load amount at each point of the distribution system as demand assumption information based on the demand pattern of new consumers and the demand pattern of existing consumers."

[0020] Also, in the present invention, there is provided "a power quality management method that manages the power quality of a distribution system using demand assumption information, which is the load amount at each point of the distribution system obtained in a demand management method, wherein in the control of an SVC that is arranged in the distribution system and adjusts the secondary side tap position according to a setting value, the demand assumption information, which is the load amount at each point of the distribution system, is used for calculating the setting value."

[0021] In the present invention, there is provided a power quality management method for predicting the power demand of a power distribution system using demand assumption information, which is the load amount at each point of the power distribution system obtained in the demand management method. The power quality management method is characterized in that, when predicting the power demand of the power distribution system, demand assumption information, which is the load amount at each point of the power distribution system, is used.

Effect of the Invention

[0022] According to a typical form of the present invention, it is possible to obtain the grasp of the effective power demand of consumers and the response characteristics to information provision, so that it is possible to obtain effects such as improvement of business performance by more appropriate guidance of consumers.

Brief Description of the Drawings

[0023]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Best Mode for Carrying Out the Invention

[0024] Hereinafter, embodiments of the present invention will be described with reference to the drawings.

Embodiment

[0025] In Embodiment 1, a power demand management system will be described. FIG. 1 shows an overall configuration example of a power demand management system 10 according to Embodiment 1 of the present invention.

[0026] A power demand management system 10 formed using a computer includes a new customer information database DB1 that holds information D1 of new customers in advance and a demand assumption information database DB2 that holds demand assumption information D2 after the implementation of a demand management plan. Also, when functionally representing the processing content of the arithmetic unit in the computer, it is configured to include at least the functions of a new customer information completion unit 30, an existing customer clustering unit 40, a demand pattern generation unit 50, and a demand volume calculation unit 60 for each location.

[0027] According to the configuration of Embodiment 1 illustrated in FIG. 1, in the power demand management system 10, demand assumption information D2 is obtained by taking into account information D1 of new customers in the information of existing customers and estimating the power demand in an assumed new power system regionally and temporally. In particular, at this time, for the information D1 of new customers, which is often lacking in available information, the information is supplemented and enriched to improve the estimation accuracy of the demand assumption information D2.

[0028] FIG. 2 shows an example of the data structure of the new customer information database DB1. The new customer information database DB1 is configured to include, as new customer information D1, a customer ID (D11), an address D12, contract information D13, a contract capacity D14, a business type D15, the number of residents D16, home appliances to be stored D17, a daytime at-home rate D18, and the like.

[0029] This information is publicly obtained by the power company from the customer, based on the timing when the customer and the power company conclude a contract, or a questionnaire to the customer. There are cases where the value is determined for each power customer with whom the power company has contracted. However, for information such as the number of residents D16, household appliances to be stored D17, and daytime home occupancy rate D18, accurate information is not always obtained from the customer.

[0030] As a result, if necessary, for the daytime home occupancy rate D18, as one of the examples of this embodiment, based on the values (%) of the statistical survey on the daytime home occupancy rate every hour by the "National Living Statistics" published by NHK, it is determined by allocation from the same contract capacity D14 and the number of residents D16. If the number of residents D18 is not obtained, it may be necessary to take measures such as determining from only the contract capacity D14. However, these measures will be carried out manually as appropriate.

[0031] As described above, not all items of the new customer information database DB1 in FIG. 2 are filled, and the current situation is that the replenishment of items has no choice but to rely on manual work.

[0032] FIG. 3 shows the processing flow in the new customer information completion unit 30. First, in processing step S301, a questionnaire is conducted for the new customer and the answers to the questionnaire questions are quantified. In this case, the questionnaire questions are determined so that the data items of the new customer information database DB1 in FIG. 2 can be obtained. However, in reality, answers corresponding to the items may not be obtained, and it is considered that unclear information is generated.

[0033] In processing step S302, the Euclidean distance of each customer in the N-dimensional space with respect to the number of questions N is calculated. Next, in processing step S303, a group is searched for such that the Euclidean distance between customers becomes small. For example, when the respondent is Mr. A, other new customers showing a similar answering tendency to Mr. A are extracted as group A. As a result, it is considered that, for example, single persons, young couple households, child-rearing families, etc. are grasped as groups showing similar answering tendencies.

[0034] Furthermore, in processing step S304, the average of the responses of each group is obtained, and the centroid vector is calculated. After extracting the customer information from the customer data, the customer information is normalized to the range of 0 to 1, and then statistical information such as the average and variance for each data in other customer information is calculated. As a result, it is assumed that even within the same group, a distribution within the group reflecting differences in working styles, income, etc. will appear.

[0035] Subsequently, in processing step S305, a feature quantity is calculated by dividing the amount of deviation from the average by the variance in specific customer information. After that, a feature vector of the customer information is obtained for each customer, and a classification tree is generated based on the feature vector. If there is unknown information among the customer information data in each customer group classified based on the classification tree, the customer information may be supplemented with information such as the average value and confidence interval from the vector components of the same classified customers.

[0036] As a result, for example, when it is determined that the probability of being a "household with two full-time co-workers" is high as a feature in the same classification tree, data can be supplemented by adding "2 people", "0%", etc. to the items of "number of residents" and "midday home occupancy rate" that were previously unknown.

[0037] FIG. 4 shows the processing of the existing customer clustering unit 40. First, in processing step S401, the feature vectors of the existing customers are calculated respectively. Although not shown in FIG. 1, it is assumed that the existing customers also separately hold data with the same configuration as in FIG. 2. Also, the calculation of the feature vector may be performed including not only the information of the existing customers but also the information of the new customers.

[0038] In processing step S402, the feature vectors of the customers are clustered. Examples of clustering methods include the Kmeans method, the Ward method, and random forest.

[0039] In processing step S403, for example, combination patterns with a reliability of 90% are extracted respectively. In processing step S404, the feature vectors of the customers are classified into a plurality of customer groups. Through the confirmation process for the completion of the classification in processing step S405, if the classification is not completed, the process moves to processing step S407 to update the values of the feature vectors of each customer and then returns to the process of processing step S404, repeating until the classification is completed.

[0040] Fig. 5 shows the processing of the demand pattern generation unit 50. In the processing of the demand pattern generation unit 50, for the customer information supplemented by the new customer information supplementation unit 30, the address D12, contract information D13, contract capacity D14, business type D15, number of residents D16, daytime home occupancy rate D18, new customer information, and meteorological information are associated, and clustering processing based on the load pattern is performed.

[0041] Then, a decision tree is created based on the supplemented customer information, namely the address D12, contract information D13, contract capacity D14, business type D15, number of residents D16, and daytime home occupancy rate D18, from the customer groups based on the load pattern obtained above. By doing so, it means that reclassification is performed based on the characteristics of the customers among similar load patterns.

[0042] For example, according to the supplemented customer information, if the probability that the household of Mr. A is a "full-time dual-income household of two people" is high, and the similar load pattern in other "full-time dual-income households of two people" is that the demand on weekdays from 8 am to around 8 pm is low and the load demand is high from 8 pm to 2 am, the load pattern of Mr. A's household can be estimated to be similar to such a similar load pattern. Note that the load pattern can be appropriately formed in units such as days, weeks, months, seasons, weekdays, and holidays.

[0043] Fig. 6 shows the processing of the demand amount calculation unit 60 at each location. In the flow of Fig. 6, first in processing step S601, an area is determined. The area corresponds to an administrative division or a map mesh. In processing step S602, consumers in the corresponding area are selected based on their addresses. In processing step S603, for the set of selected consumers, the estimated demand and the actual demand amount are calculated as the load at each time section. Note that information representing a consumer may be given as a location. In this case, the location may be the latitude and longitude, or the total load within the range obtained by dividing the map into 10 km x 10 km meshes.

[0044] In processing step S604, the load factor in the area is set. Based on the load factor, in processing step S605, the active power P and the reactive power Q are calculated for each time section from the demand amount. Moreover, the difference between the total load and the maximum allowable load at each location is the load margin. The load margin can be calculated for the active power P and the reactive power Q if the power factor is given. Note that it is possible to grasp the active power and reactive power of power generation by combining the installation capacity data of distributed power sources such as distributed solar power generation connected to the distribution line and the meteorological data in the area. Also, even when power generation and load exist behind the smart meter data and meteorological data exists, it is possible to grasp the actual load by assuming the power generation amount from the meteorological data and adding it to the smart meter data.

[0045] Fig. 7 shows an example of the data structure of the demand assumption information database DB2. The demand assumption information D2 is composed of an area ID D21, the active power D22 for each time, the reactive power D23, the connected distribution line D24, etc.

[0046] According to the demand assumption information D2 obtained by the process of FIG. 6 and stored in the demand assumption information database DB2 of FIG. 7, it is possible to grasp the distribution of loads (active power D22, reactive power D23) on a new power system configuration in terms of geography (area ID, D21) including existing demand customers and new demand customers. Further, this load distribution can be grasped not only geographically but also in terms of time series (vertical axes of active power D22 and reactive power D23) such as day, week, and month. Furthermore, since the geographical load distribution grasped here includes address information, it will be grasped in relation to equipment positions such as switches and circuit breakers on the distribution line and information on the branches of the distribution line in the power system (connected distribution line D24).

[0047] According to the power demand management system 10 illustrated in the first embodiment, by complementing the shortage information about new demand customers, the accuracy of customer data is improved, and as a result, it becomes possible to improve the accuracy of the finally obtained demand assumption information D2.

Embodiment

[0048] In the second embodiment, a power quality management system in which the demand assumption information D2 estimated by the power demand management system of the first embodiment is reflected in the monitoring and control of the power system will be described. Since the demand assumption information D2 for each distribution line can be accurately assumed by using the power quality management system, it can be applied to power quality management.

[0049] FIG. 8 shows an example of applying to an SVR (Step Voltage Regulator) as an application example of the power quality management system. On the left side of FIG. 8, the power demand management system 10 (load management unit) and the demand assumption information D2 estimated here are shown. On the right side of FIG. 8, as an example of the power quality management system, it shows that the function of the setting unit 30 for determining the setting value of the SVR is realized. Note that DB3 is an existing customer information database that stores the existing customer information D3.

[0050] Here, briefly describe the necessity of installing SVR. First, the voltage of the distribution line varies depending on the supply point and time zone due to the line voltage drop caused by the load power. The high-voltage voltage regulator is installed in the middle of the line for the purpose of adjusting the voltage of the high-voltage distribution line so that it falls within the specified voltage range. As a representative example of this, SVR is generally used. SVR uses a single-winding transformer, and voltage control is performed by automatically switching its taps. The output voltage of the SVR is controlled based on the reference voltage and by setting the LDC operating conditions. Based on the reference voltage, an LDC compensation value (the line voltage drop value from the SVR installation point to the load center point thereafter), which is proportional to the magnitude of the load current, is added to the value. Note that LDC (Line Drop Compensation) is a method of keeping the voltage at a certain point constant.

[0051] In the voltage calculation unit 31 within the setting unit 30 in FIG. 8, based on the demand assumption information D2 obtained by the power demand management system 10 (load management unit), a power flow calculation is used to perform a voltage calculation of the distribution system from the network topology and line impedance related to the distribution line stored in the distribution system information. By this calculation, the current and voltage values at each node (point) of the distribution line can be calculated for each time cross-section.

[0052] Based on this result, in the voltage regulator setting unit 32 within the setting unit 30, a setting value 33 (Vref, R, X) for estimating the load center point voltage of the distribution line is calculated. This calculation method is LDC. The SVC installed on the distribution line performs tap control according to the setting value 33 to control the voltage of the distribution line.

[0053] Note that this setting value 33 is used for a period determined for each distribution line. The period may be commonly used for several years such as one year, or may be used separately for weekdays and holidays. Furthermore, it is possible to use it separately for each season, or separately for daytime and nighttime.

[0054] The values of such setting values can be calculated by thoroughly grasping the power usage behavior of consumers based on smart meter data or the like. By the power demand management system 10, since the behavior of new consumers can be accurately assumed, it is possible to calculate the setting values more accurately at various time intervals.

[0055] FIG. 9 shows an example of general processing content cases for more diverse applications of the power quality management system, and shows the processing flow when using the demand assumption information D2 estimated by the power demand management system not only for voltage but also for general power analysis. The flow in FIG. 9 shows an example of analysis processing such as voltage estimation performed by the voltage calculation unit 31.

[0056] An example of a distribution system, which is a general monitoring target, is shown at the upper part of the speech balloon in FIG. 9. The distribution line is appropriately branched and arranged from the circuit breaker FCB at the substation exit via a switch with sensor DS, a manual switch DSh, etc. In order to perform quality management of such a distribution system, it is necessary to obtain voltage information etc. at various locations. Currently, measurement data from the switch with sensor DS, smart meter (measurement data: power consumption amount (30-minute value)) data, and customer load data are available.

[0057] However, in any case, their spread is not sufficient in number, and they are only obtained within the time span required for quality management, and there are also problems with data accuracy. In terms of data accuracy, customer load data is the most suitable, but there is also a problem with the spread, and it is inevitable that non-measurement sections will occur on the distribution system.

[0058] Regarding these current measurement situations, according to the demand assumption information D2 by the power demand management system 10, since the demand data for each distribution line can be accurately assumed regardless of the location, it can be applied to planning support. Conventionally, it was a simple monitoring based on the measurement values in a limited number of switches with sensors DS, but by using the demand assumption information D2 by the power demand management system 10, it will be possible to further enhance the monitoring technology in the future.

[0059] On the other hand, for the advancement of power distribution system monitoring technology, there are the following needs according to the operation purpose. First, regarding overload, it is necessary to grasp in detail the maximum value and the occurrence time so that the allowable value of equipment (distribution lines, switches) does not deviate. Also, regarding equipment utilization rate, it is necessary to manage in detail the daily power consumption and use it when considering equipment capacity during equipment planning. Further, regarding voltage distribution, it is necessary to grasp in detail when and where voltage rise occurs due to the connection of renewable energy such as PV.

[0060] Regarding the above needs, in the current power distribution system, due to its wide area and limited sensor installation locations, it is impossible to accurately grasp the power flow and voltage conditions in branch sections and low-voltage systems, which are non-measured locations.

[0061] Therefore, in Example 2, the demand assumption information D2 is used, and a method using clustering is applied to extract features for enhanced monitoring for each operation purpose. Clustering is a technique for grouping individuals with multiple evaluation values into the same group based on proximity of evaluation values and extracting features.

[0062] At the power distribution system site, attention is paid to overload, equipment utilization rate, and voltage distribution. Therefore, in FIG. 9 illustrating the processing in the voltage calculation unit 31, in the first processing step S801, a load model is generated according to the management items. These include generating a model for the maximum value and occurrence time regarding overload in power flow management (processing step S801a), generating a model for the monthly power consumption regarding equipment utilization rate in power flow management (processing step S801b), and generating a model for obtaining the voltage distribution for each time regarding voltage management (processing step S801c).

[0063] Regarding these models, in the power distribution system shown in the upper part of the callout in FIG. 9, sensor installation sections and non-measured sections will occur. Therefore, in processing step S802, the load model in the sensor installation section is estimated, and in processing step S803, the load model in the non-measured section is estimated and corrected to obtain a corrected load model.

[0064] In this estimation, it is advisable to use the demand assumption information D2 by the power demand management system 10. For example, in the power distribution system shown in the upper part of the balloon in FIG. 9, when the combination is two sensor installation sections and two non-measured sections, if the estimation of the sensor installation section can be accurately performed from the sensor detection values, it is performed using the measurement data. When there is a problem that the sensor accuracy is low although it is in the sensor installation section, or when it is a non-measured section, it is advisable to use the demand assumption information D2 by the power demand management system 10.

[0065] By geographically connecting these sensor installation sections and non-measured sections, the load state at each point from the substation outlet to the end of the distribution line can be obtained as a geographical distribution. Also, as shown in the lower part of the balloon in FIG. 9, for example, it can be obtained as a time-series change of the power amount at a specific point on the distribution line. In this example, it shows that it is obtained by comparing the case obtained by the measurement sensor and the case using the demand assumption information D2 by the power demand management system 10. These processes are performed in processing step S804, and for example, the voltage and current of the entire section are obtained with the corrected load model.

[0066] In processing step S805, the analysis result is displayed and output on the display. The display format at this time is preferably the display format described in the balloon in FIG. 9.

[0067] According to the above-described Example 2, the voltage estimation that accurately grasps the load distribution for each section is obtained, and the effect that the deviation from the true value is reduced is obtained.

Example

[0068] In Example 2, a power quality management system that reflects the demand assumption information D2 estimated by the power demand management system of Example 1 in demand management in a VPP (Virtual Power Plant) or DR (Demand Response) of the power system will be described.

[0069] Note that VPP (Virtual Power Plant) means that the owners of energy resources on the customer side, power generation facilities directly connected to the power grid, energy storage facilities, or third parties control their energy resources to provide functions equivalent to those of a power plant. DR (Demand Response) means that the owners of energy resources on the customer side or third parties control their energy resources to change the power demand pattern.

[0070] Generally, in the backbone power grid, various information such as temperature and solar radiation is utilized to conduct detailed demand forecasting. On the other hand, in the distribution power grid, only the sensor information installed in the distribution power grid is utilized for simple forecasting. In the future, in order to further improve the efficiency of facility planning, it is necessary to enhance the demand forecasting for distribution lines.

[0071] Different from the backbone power grid, the demand of the distribution power grid is characterized by regional differences according to distribution lines and sections, the contract type and capacity of customers, and lifestyle (presence or absence of being at home). Therefore, for the distribution line prediction method, a prediction method using detailed smart meter data and data analysis results such as temperature is required. It is possible to collect multiple distribution line loads and apply big data analysis (such as clustering technology and regression analysis) to analyze the influencing factors of the demand curve and smart meter data fluctuations.

[0072] Figure 10 shows an example of applying a distribution line demand forecasting system for forecasting the demand of a distribution line as an example of a power quality management system. The distribution line demand forecasting system is equipped with a distribution line information database DB5 and an estimated parameter database DB6. The existing distribution line information database DB5 includes smart meter data, sensor measurement values, average temperature, etc. However, in Example 3 of the present invention, it is better to further hold the demand assumption information D2 estimated by the power demand management system of Example 1.

[0073] Furthermore, in the processing of the distribution line demand prediction system, specifically, it collects smart meter data of general lighting customers, analyzes the demand curve, and clarifies the influencing factors such as region, weather information, contract type and capacity of customers. Finally, it examines and evaluates a method for predicting the future demand curve using the following parameters (day-of-week classification, time classification, temperature, etc.). Based on such predictions, it formulates a plan on whether to implement demand response or not.

[0074] In the example of Fig. 10, the distribution line demand prediction is composed of a part that is started daily to perform statistical processing (cluster analysis and regression analysis) of the previous day's smart meter data and machine-learn the parameters for demand prediction, and a part that is started when demand prediction is required to perform the demand prediction of the distribution line.

[0075] In processing step S101, using the information in the distribution line information database DB5, analyze the demand curve (cluster analysis) according to the time classification of the previous day. Similarly, in processing step S102, using the information in the distribution line information database DB5, analyze the demand curve (cluster analysis) according to the day-of-week classification of the previous day. As a result, for example, regarding the power consumption every 30 minutes from 0:00 to 24:00, the characteristic movements for each day of the week are measured for the power consumption around 8:00 am and around 8:00 pm, and these are classified as new events respectively.

[0076] Furthermore, in processing step S103, using the information in the distribution line information database DB5, obtain the correlation between the average temperature of the previous day and the power consumption by regression analysis. Thereby, the relationship between temperature and power becomes clear, and it becomes possible to estimate the power consumption with the influence of temperature as another factor. In processing step S104, considering these clusterings and correlations, obtain the estimated parameters as feature quantities and accumulate them in the estimated parameter database DB6.

[0077] Next, in the demand forecasting process, when demand forecasting is activated in processing step S105, in processing step S106, a distribution line to be used for demand forecasting is determined. In processing step S107, demand forecasting is performed and output using the parameters stored in the estimation parameter database DB6. Note that the conventional distribution line demand forecasting method performs demand forecasting based on the growth rate of past sensor measurement values.

[0078] In the present invention, distribution line demand forecasting is performed based on the results of analyzing the factors affecting the variation of the demand curve obtained from the smart meter data and the demand assumption information D2 estimated by the power demand management system. By using smart meter data, time division, day-of-week division, contract kW, and temperature as input data, it becomes possible to improve the accuracy of distribution system line demand forecasting that captures the individual characteristics of consumers.

[0079] FIG. 11 shows the relationship between the demand (predicted) and the induced amount (predicted) considering non-linearity. Each axis in FIG. 11 represents time, temperature, and power consumption. Here, as a non-linear function, if it is expressed as a polynomial having a product with a maximum order of 5 in the time direction and a maximum order of 2 in the temperature direction within a one-day range, the behavior of the daily power demand at various temperatures can be simulated. Then, the partial differential coefficient is obtained at the cross-section of each time and each temperature. This differential coefficient is regarded as the sensitivity to temperature and time and is held as a table to calculate the demand and the induced amount.

[0080] Note that, in order to model human behaviors such as using electricity and not using electricity, the sigmoid curve, which is one of the cumulative distribution functions, is used. The sigmoid curve of equation (1) is a horizontally stretched S-shaped function, which takes a value of 0 at the limit of -∞ and a value of 1 at the limit of +∞. In physics, it is known that the change in the vertical electron spin when an external magnetic field is applied to a magnetic substance, the growth curve in economics, and the mortality curve in biology can also be approximated by the sigmoid curve.

[0081]

Equation

[0082] In addition, since this sigmoid curve can be expressed by equation (2) as a function of the original sigmoid curve through differentiation, it has the characteristic that it is easy to calculate without the need for numerical calculations for differential operations in computer simulations.

[0083]

Number

[0084] Note that the sigmoid curve can be translated parallel to the x-axis or the magnitude of the slope at the inflection point can be changed by the value of a in the equation. Mathematically, differentiating the cumulative distribution function results in the probability density function. In the present invention, by assigning the aforementioned partial differential coefficient to a in the equation, the sigmoid curve (cumulative distribution function) and the probability density function can be obtained by differentiating the sigmoid curve. Thereby, the response probability to demand response can be expressed.

Explanation of Signs

[0085] 10: Demand Management System DB1: New Customer Information Database 30: New Customer Information Completion Section 40: Existing Customer Clustering Section 50: Demand Pattern Generation Section 60: Demand Quantity Calculation Section for Each Location DB2: Demand Forecast Information Database

Claims

1. For the information of new customers composed of multiple items, there is a new customer information completion unit that complements the information of unclear items; for the information of existing customers composed of multiple items, there is an existing customer clustering unit that extracts customers with information similar to that of existing customers and classifies existing customers; from the demand patterns in the set of customers with the similar information, there is a demand pattern generation unit that generates the demand patterns of new customers; and based on the demand patterns of new customers and existing customers, there is a demand quantity calculation unit for each location that calculates the load quantity at each location of the power distribution system as demand assumption information. The new customer information completion unit numerically values the answers to each question of the questionnaire implemented for multiple new customers, calculates the Euclidean distance between each new customer in a dimensional space with the same number of dimensions as the number of questions in the questionnaire, searches for groups with a smaller Euclidean distance between each new customer, obtains the average of the questionnaire answers of each group to calculate the centroid vector, extracts the characteristics of each group based on the centroid vector, and when there is information on unclear items in the information of any new customer classified into each group according to the characteristics, complements the information on the unclear items from the information of other new customers in each group. A demand management system characterized by this.

2. The demand management system according to claim 1, wherein the existing customer clustering unit uses the K-means method. A demand management system characterized by this.

3. A power quality management system that manages the power quality of a power distribution system using the demand assumption information, which is the load quantity at each location of the power distribution system obtained in the demand management system according to any one of claim 1 or claim 2, The power quality management system includes a control device for an SVC that is arranged in the power distribution system and adjusts the secondary side tap position according to a setting value, and uses the demand assumption information, which is the load quantity at each location of the power distribution system, for calculating the setting value. A power quality management system characterized by this.

4. A power quality management system that performs power demand prediction for a power distribution system using the demand assumption information, which is the load amount at each point of the power distribution system obtained in the demand management system according to any one of claims 1 or 2. The power quality management system is characterized in that, when performing power demand prediction for the power distribution system, the power quality management system uses the demand assumption information, which is the load amount at each point of the power distribution system.

5. The power quality management system according to claim 4. When performing power demand prediction for the power distribution system, the power quality management system approximates the factors of power demand by a non-linear function, and performs multi-dimensional sensitivity analysis using the partial differential coefficients at each coordinate of the non-linear function, and has means for calculating the customer response probability after determining a probability density function based on the partial differential coefficients. The power quality management system is characterized by this.

6. The power quality management system according to claim 5. The non-linear function has a product of terms with a maximum order of 5 in the time direction and a maximum order of 2 in the temperature direction. As the probability density function projected from the partial differential coefficients of the non-linear function, a sigmoid function is used. The power quality management system is characterized by this.

7. Regarding the information of new customers composed of multiple items, complement the information of unknown items, extract customers with information similar to the information of existing customers regarding the information of existing customers composed of multiple items, classify the existing customers, generate the demand pattern of new customers from the demand patterns in the set of customers with the similar information, and calculate the load amount at each point of the power distribution system as demand assumption information based on the demand pattern of new customers and the demand pattern of existing customers. In the complementation of information on the unknown items, the responses to each question of the questionnaire implemented for a plurality of new customers are quantified, the Euclidean distance between each new customer in a dimensional space equal to the number of questions of the questionnaire is calculated, a group with a smaller Euclidean distance between each new customer is searched for, the average of the questionnaire responses of each group is obtained to calculate a centroid vector, the characteristics of each group are extracted based on the centroid vector, and when there is information on an unknown item in the information of any new customer classified into each group according to the characteristics, the information on the unknown item is complemented from the information of other new customers in each group. A demand management method characterized by this.

8. The demand management method according to claim 7, A demand management method characterized by using the K-means method when classifying existing customers.

9. A power quality management method for managing the power quality of a distribution system using the demand assumption information, which is the load amount at each point of the distribution system obtained in the demand management method according to any one of claims 7 or 8, In the control of an SVC arranged in a distribution system and adjusting the secondary side tap position according to a setting value, a power quality management method characterized by using the demand assumption information, which is the load amount at each point of the distribution system, in the calculation of the setting value.

10. A power quality management method for performing power demand prediction of a distribution system using the demand assumption information, which is the load amount at each point of the distribution system obtained in the demand management method according to any one of claims 7 or 8, A power quality management method characterized by using the demand assumption information, which is the load amount at each point of the distribution system, when performing power demand prediction of the distribution system.

11. The power quality management method according to claim 10, When performing power demand prediction for a power distribution system, approximate the factors of power demand with a non-linear function, perform multi-dimensional sensitivity analysis using the partial differential coefficients at each coordinate of the non-linear function, determine a probability density function based on the partial differential coefficients, and calculate a customer response probability. A power quality management method characterized by this.

12. The power quality management method according to claim 11, The non-linear function has a product of terms with a maximum order of 5 in the time direction and a maximum order of 2 in the temperature direction, and a sigmoid function is used as the probability density function projected from the partial differential coefficients of the non-linear function. A power quality management method characterized by this.

Images