Abnormal pricing detection method based on support vector regression and influence difference

Abstract

The invention discloses an abnormal pricing detection method based on support vector regression and an influence difference value, and is applied to the technical field of intelligent community coordinated scheduling. Comprising the following steps: establishing an electric energy consumption model and a pricing model related to intelligent home scheduling in a community; modeling is carried out on a centralized smart home scheduling problem, the centralized smart home scheduling problem is solved based on an electric energy consumption model and a pricing model, and dynamic planning and scheduling are carried out on household appliances of clients, so that the personal currency cost of the clients is minimized; calculating a predicted guidance price curve by using a support vector regression algorithm based on a dynamic planning scheduling result; and comparing the guidance price curve received by the intelligent electric meter of the client with the predicted guidance price curve, and detecting whether abnormal pricing exists in the guidance price curve received by the intelligent electric meter. The invention provides a single event detection technology based on support vector regression and influence difference, which can effectively detect the abnormal pricing event of the electricity price.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent community coordination and scheduling, and more specifically to an abnormal pricing detection method based on support vector regression and influence difference. Background Art

[0002] Smart home controls the final use of the power grid and is an important part of the smart grid infrastructure. The scheduling technology of smart home plays a key role in the smart home system. It controls the energy use of customers. Its basic principle is to reduce energy consumption during high electricity price periods and transfer it to low electricity price periods. In this way, the average cost of energy consumption of customers is reduced, and the energy load of the power grid is balanced within a certain time range. The electricity price curve based on the scheduling of smart home can guide the energy use of users, so that the energy load of the community is balanced. There are two pricing modes in the current smart grid. The first is real-time pricing. According to the energy consumed by customers in a time window in the past, each customer who consumes energy in the time window is charged. The real-time pricing is determined according to the actual energy consumption; the second is guided pricing. Since the smart home scheduler decides in advance when to start each device, it needs to know the future electricity price for scheduling planning. However, the electricity price depends on the actual energy consumption in the future. Therefore, utility companies must predict future loads, design a predicted pricing curve, and use it to guide the cheap time period of the smart home scheduler. This predictive pricing is called guided pricing.

[0003] For the guidance pricing, since the guidance pricing needs to be sent to the customer in advance during use, if there is a specific network attack to modify the guidance pricing curve, it will cause the customer's smart home scheduling plan to not meet the requirements and increase the peak energy consumption of the local power system. If it is possible to identify whether the guidance pricing received by the customer has abnormal pricing behavior, the above losses can be avoided. However, there is currently no detection method for whether the guidance pricing is abnormal. Therefore, how to provide an abnormal pricing detection method based on support vector regression and impact difference is a problem that technicians in this field urgently need to solve. Summary of the invention

[0004] In view of this, the present invention provides an abnormal pricing detection method based on support vector regression and influence difference, which is used to detect abnormal pricing events of electricity price manipulation.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] An abnormal pricing detection method based on support vector regression and influence difference includes the following steps:

[0007] S1. Establish the energy consumption model and monetary cost model related to smart home scheduling in the community;

[0008] S2. Model the centralized smart home scheduling problem, solve the centralized smart home scheduling problem based on the power consumption model and the monetary cost model, and dynamically plan and schedule each customer's home appliances to minimize the customer's personal monetary cost;

[0009] S3. Based on the dynamic programming scheduling results, the support vector regression algorithm is used to calculate the predicted guidance price curve;

[0010] S4. Compare the guidance price curve received by the customer's smart meter with the predicted guidance price curve to detect whether there is abnormal pricing in the guidance price curve received by the smart meter.

[0011] Optionally, S1 is:

[0012] S11. For all users in the community, calculate the energy consumption of household appliances of each user under the smart home scheduling to obtain an energy consumption model;

[0013] S12. Based on the energy consumption of each user's household appliances and using real-time electricity prices, calculate the monetary cost of each user in a certain period of time to obtain a monetary cost model.

[0014] Optionally, the power consumption model is specifically:

[0015]

[0016] In the formula, For all customers in the community, L h is the total electricity consumption of the community in time period h, is the power consumption of all manually controlled appliances in customer n’s home during time period h, is the power consumption of all automatically controlled appliances in customer n’s home during time period h, represents all manually controlled appliances in customer n's home, represents all the automatically controlled appliances in customer n's home, t m,h represents the actual execution time of appliance m in time period h, x m,h represents the working power of household appliance m in time period h, where For all the home appliances in the customer's home,

[0017] Optionally, the monetary cost model is as follows:

[0018] C h =a h L h2

[0019]

[0020] In the formula, C n,h is the monetary cost of customer n in time period h, is the power consumption of all manually controlled appliances in customer n’s home during time period h, is the power consumption of all automatically controlled appliances in customer n’s home during time period h, C h is the total monetary cost of the community in time period h, L h is the total electricity consumption of the community in time period h, a h is the pricing parameter.

[0021] Optionally, S2 is specifically:

[0022] S21, each customer solves the centralized smart home scheduling problem based on the power consumption model and the monetary cost model to obtain the energy consumption of all other customers in the community;

[0023] S22, each customer updates its own power consumption based on the dynamic planning scheduling results and shares it with all other customers;

[0024] S23, each customer updates the dynamic planning schedule of home appliances based on the energy consumption of all other customers;

[0025] S24. Repeat S21-S23 until convergence to obtain the final dynamic planning scheduling result.

[0026] Optional, S3 is:

[0027] S31. Obtain the historical guidance price in the last T days and h time period;

[0028] S32. Derive the predicted price based on the historical guidance price in the recent T days and h time period.

[0029] Optionally, S4 is specifically:

[0030] S41. Calculate a predicted bill and a predicted PAR value based on the predicted guide price curve, wherein the PAR value is a peak-to-average ratio;

[0031] S42, calculating the actual bill and the actual PAR value based on the received guidance price curve;

[0032] S43, calculating the bill growth rate and the PAR value growth rate;

[0033] S44. If the bill growth rate is greater than the preset tolerable threshold or the PAR value growth rate is greater than the preset tolerable threshold, it is determined that abnormal pricing exists and an alarm is issued.

[0034] It can be seen from the above technical solution that compared with the prior art, the present invention provides an abnormal pricing detection method based on support vector regression and influence difference, which has the following beneficial effects: the present invention evaluates the vulnerability of electricity price transmission in smart home systems, and considers the two closely related pricing network attacks that manipulate smart meter guidance electricity prices, namely, bill reduction and peak energy load, so as to effectively detect abnormal pricing events of electricity price manipulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0036] Figure 1 is a flow chart of the abnormal pricing detection method of the present invention;

[0037] Figure 2 This is a flow chart of the dynamic programming scheduling algorithm of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] The embodiment of the present invention discloses an abnormal pricing detection method based on support vector regression and influence difference, such as Figure 1 As shown, the following steps are included:

[0040] S1. Establish the energy consumption model and monetary cost model related to smart home scheduling in the community;

[0041] S2. Model the centralized smart home scheduling problem, solve the centralized smart home scheduling problem based on the power consumption model and the monetary cost model, and dynamically plan and schedule each customer's home appliances to minimize the customer's personal monetary cost;

[0042] S3. Based on the dynamic programming scheduling results, the support vector regression algorithm is used to calculate the predicted guidance price curve;

[0043] S4. Compare the guidance price curve received by the customer's smart meter with the predicted guidance price curve to detect whether there is abnormal pricing in the guidance price curve received by the smart meter.

[0044] Furthermore, S1 is specifically:

[0045] S11. For all users in the community, calculate the energy consumption of household appliances of each user under the smart home scheduling to obtain an energy consumption model;

[0046] S12. Based on the energy consumption of each user's household appliances and using real-time electricity prices, calculate the monetary cost of each user in a certain period of time to obtain a monetary cost model.

[0047] Furthermore, the power consumption model is specifically as follows:

[0048]

[0049] In the formula, For all customers in the community, L h is the total electricity consumption of the community in time period h, is the power consumption of all manually controlled appliances in customer n’s home during time period h, is the power consumption of all automatically controlled appliances in customer n’s home during time period h, represents all manually controlled appliances in customer n's home, represents all the automatically controlled appliances in customer n's home, t m,h represents the actual execution time of appliance m in time period h, x m,h represents the working power of household appliance m in time period h, where For all the home appliances in the customer's home,

[0050] Furthermore, in the embodiment of the present invention, for each automatically controlled household appliance, its working power in the time period h is Subject to the following constraints:

[0051] For a specific task, the total energy consumption of appliance m is expressed as E m Indicates that E m is fixed and can be calculated in advance. It is equal to the sum of energy consumption within the time range, so:

[0052]

[0053] Appliance m should start at the requested start time α no earlier than m and no later than the deadline β m Complete the task, namely:

[0054]

[0055] In the h time period, the energy consumption of user n is equal to that of all household appliances The total energy consumption is such that:

[0056]

[0057] Furthermore, the monetary cost model is as follows:

[0058] C h =a h L h 2

[0059]

[0060] In the formula, C n,h is the monetary cost of customer n in time period h, is the power consumption of all manually controlled appliances in customer n’s home during time period h, is the power consumption of all automatically controlled appliances in customer n’s home during time period h, C h is the total monetary cost of the community in time period h, L h is the total electricity consumption of the community in time period h, a h is the pricing parameter.

[0061] Further, such as Figure 2 As shown, S2 is specifically:

[0062] S21, each customer solves the centralized smart home scheduling problem based on the power consumption model and the monetary cost model to obtain the energy consumption of all other customers in the community;

[0063] S22, each customer updates its own power consumption based on the dynamic planning scheduling results and shares it with all other customers;

[0064] S23, each customer updates the dynamic planning schedule of home appliances based on the energy consumption of all other customers;

[0065] S24. Repeat S21-S23 until convergence to obtain the final dynamic planning scheduling result.

[0066] When the user's dynamic planning schedule no longer changes, or the change in the dynamic planning schedule is lower than the set threshold, it is judged to be converged.

[0067] Furthermore, in the embodiment of the present invention, S21 is specifically:

[0068]

[0069] Since the monetary cost of each customer n depends on the total energy load of the community including all other customers, the dispatch of one customer will have an impact on other customers. The monetary cost of each customer n can be divided into two parts, which are:

[0070]

[0071] In the formula, l -n,h is the energy load of the entire community except the energy consumption of customer n in time slot h, and we have:

[0072]

[0073] In solving this problem, the goal of each customer is to minimize the total payment amount by scheduling the energy consumption of the automatically controlled appliances. When no one can further reduce their own monetary cost without changing other customers, the Nash equilibrium is reached. To solve this problem, this embodiment adopts a decentralized technology, which is an iterative algorithm. In each iteration, each customer uses an algorithm based on dynamic programming to solve P1, while assuming that the energy consumption of all other users, i.e., l -n,h is fixed. After each iteration, each customer obtains new energy consumption information from all other customers according to the updated scheduling scheme. Based on the updated energy consumption of each customer, update l -n,h , and according to the updated l -n,h , dynamically plan and schedule each customer's home appliance. This process is iterated continuously until convergence.

[0074] Furthermore, S3 is specifically:

[0075] S31. Obtain the historical guidance price in the last T days and h time period;

[0076] S32. Derive the predicted price based on the historical guidance price in the recent T days and h time period.

[0077] Furthermore, in the embodiment of the present invention, the historical guidance electricity price for the last T days is represented by the following matrix:

[0078]

[0079] Use a h =[a 1,h ,a 2,h ,…,a T,h ] represents the guidance electricity price for the last T days and h periods;

[0080] Calculate the function f(a h ), by a h Derived h time slot (with The forecast guidance electricity price is as follows:

[0081]

[0082] Given the historical guidance electricity price, the function f(a h ):

[0083]

[0084] In the formula, K(a i ,a j ) is defined as the kernel function, a i 、a j They represent the guidance electricity prices for time periods i and j respectively, and u i 、u j Respectively represent the optimization coefficients of time periods i and j. In the present invention, radial basis kernel function is used to obtain u by solving problem P2. i , function f(a h ):

[0085] K(a i ,a j )=exp(-γ×||a i -a j || 2 )

[0086] Among them, γ is a free parameter.

[0087] In calculating the function f(a h ) and then, according to Derive the predicted guidance electricity price.

[0088] Furthermore, S4 is specifically:

[0089] S41. Calculate a predicted bill and a predicted PAR value based on the predicted guide price curve, wherein the PAR value is a peak-to-average ratio;

[0090] S42, calculating the actual bill and the actual PAR value based on the received guidance price curve;

[0091] S43, calculating the bill growth rate and the PAR value growth rate;

[0092] S44. If the bill growth rate is greater than the preset tolerable threshold or the PAR value growth rate is greater than the preset tolerable threshold, it is determined that abnormal pricing exists and an alarm is issued.

[0093] Furthermore, in an embodiment of the present invention, the maximum tolerable impact difference is a 2% increase in the peak-to-average ratio and a 5% increase in the bill; given the predicted guideline pricing curve and the received guideline pricing curve, a smart home scheduling simulation can be performed to calculate the average bill and the peak-to-average ratio PAR for each guideline pricing curve, and then the difference between them is calculated as the actual impact difference. When the actual impact difference is greater than the maximum tolerable impact difference, a potential pricing attack is discovered and an alarm signal is sent to the utility for further inspection.

[0094] In calculating the predicted guidance price curve a p After that, a smart home scheduling simulation can be performed to calculate the bill B p and PAR value P p ; By calculating the bill B and PAR value P using the received guideline price curve a, the bill growth rate is calculated as The PAR growth rate was calculated as If ΔB>δ B Or ΔP>δ P , an alarm will be issued,δ B and δ P They are the tolerable thresholds of bill growth rate and PAR value growth rate respectively.

[0095] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0096] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An abnormal pricing detection method based on support vector regression and influence difference, characterized in that: The following steps are involved: S1. Establish the energy consumption model and monetary cost model related to smart home scheduling in the community; S2. Model the centralized smart home scheduling problem, solve the centralized smart home scheduling problem based on the power consumption model and the monetary cost model, and dynamically plan and schedule each customer's home appliances to minimize the customer's personal monetary cost; S3. Based on the dynamic programming scheduling results, the support vector regression algorithm is used to calculate the predicted guidance price curve; S4. Compare the guidance price curve received by the customer's smart meter with the predicted guidance price curve to detect whether there is abnormal pricing in the guidance price curve received by the smart meter.

2. The abnormal pricing detection method based on support vector regression and influence difference according to claim 1, characterized in that: S1 is specifically: S11. For all users in the community, calculate the energy consumption of household appliances of each user under the smart home scheduling to obtain an energy consumption model; S12. Based on the energy consumption of each user's household appliances and using real-time electricity prices, calculate the monetary cost of each user in a certain period of time to obtain a monetary cost model.

3. The abnormal pricing detection method based on support vector regression and influence difference according to claim 2, characterized in that: The specific energy consumption model is: In the formula, For all customers in the community, L h is the total electricity consumption of the community in time period h, is the power consumption of all manually controlled appliances in customer n’s home during time period h, is the power consumption of all automatically controlled appliances in customer n’s home during time period h, represents all manually controlled appliances in customer n's home, represents all the automatically controlled appliances in customer n's home, t m,h represents the actual execution time of appliance m in time period h, x m,h represents the working power of household appliance m in time period h, where For all the home appliances in the customer's home, 4. The abnormal pricing detection method based on support vector regression and influence difference according to claim 2 is characterized in that: The monetary cost model is as follows: C h =a h L h 2 In the formula, C n,h is the monetary cost of customer n in time period h, is the power consumption of all manually controlled appliances in customer n’s home during time period h, is the power consumption of all automatically controlled appliances in customer n’s home during time period h, C h is the total monetary cost of the community in time period h, L h is the total electricity consumption of the community in time period h, a h is the pricing parameter.

5. The abnormal pricing detection method based on support vector regression and influence difference according to claim 1, characterized in that: S2 is specifically: S21, each customer solves the centralized smart home scheduling problem based on the power consumption model and the monetary cost model to obtain the energy consumption of all other customers in the community; S22, each customer updates its own power consumption based on the dynamic planning scheduling results and shares it with all other customers; S23, each customer updates the dynamic planning schedule of home appliances based on the energy consumption of all other customers; S24. Repeat S21-S23 until convergence to obtain the final dynamic planning scheduling result.

6. The abnormal pricing detection method based on support vector regression and influence difference according to claim 1, characterized in that: S3 is specifically: S31. Obtain the historical guidance price in the last T days and h time period; S32. Derive the predicted price based on the historical guidance price in the recent T days and h time period.

7. The abnormal pricing detection method based on support vector regression and influence difference according to claim 1, characterized in that: S4 is specifically: S41. Calculate a predicted bill and a predicted PAR value based on the predicted guide price curve, wherein the PAR value is a peak-to-average ratio; S42, calculating the actual bill and the actual PAR value based on the received guidance price curve; S43, calculating the bill growth rate and the PAR value growth rate; S44. If the bill growth rate is greater than the preset tolerable threshold or the PAR value growth rate is greater than the preset tolerable threshold, it is determined that abnormal pricing exists and an alarm is issued.

Images