Multi-energy complementary building room temperature control method in severe cold area

Abstract

The invention discloses a multi-energy complementary building room temperature control method in severe cold areas, and relates to the field of room temperature control. The invention aims to solve the problem that the indoor temperature regulation and control efficiency of the existing building is low. The method comprises the following steps: acquiring a building outdoor temperature sequence and a corresponding indoor temperature sequence, setting a label for the building outdoor temperature, and forming a training set by using the label, the outdoor temperature and the corresponding indoor temperature; and training the NFIN network by using the training set to obtain a trained NFIN network. Obtaining a current outdoor temperature and a corresponding indoor temperature, inputting the outdoor temperature and the indoor temperature into the trained NFIN network, and obtaining a result whether the current indoor temperature is appropriate or not; if the current indoor temperature is too high, the indoor temperature is reduced; if the current indoor temperature is low, the indoor temperature is increased; and if the current indoor temperature is appropriate, adjustment is not carried out. The method is used for automatically controlling the indoor temperature of the building.

Description

Technical Field

[0001] The invention relates to the field of room temperature control, and in particular to a room temperature control method for a multi-energy complementary building in severe cold regions. Background Art

[0002] With the advancement of economic development and urbanization, my country's annual new building area reaches 20×108m 2 , the proportion of building energy consumption in the total social energy consumption has reached 28%, and urban heating alone accounts for more than 25% of building energy consumption. Although certain measures have been taken, the heat consumption per unit area of ​​buildings in my country is still 1 to 2 times higher than that of some developed countries with similar climatic conditions. In my country's centralized heating system, the energy-saving potential of users' behavior is 18% to 26%, which is the largest energy-saving potential. In the context of people's increasing requirements for indoor thermal comfort, automatic adjustment and control of indoor temperature is one of the ways to solve the energy waste caused by excessively high room temperature, and it is also an important measure to ensure user comfort.

[0003] At present, indoor temperature control of buildings is mainly achieved by multi-layer back propagation neural network (BPNN), which has powerful nonlinear mapping ability and self-learning ability to achieve adaptive temperature control. However, the training time of multi-layer back propagation neural network (BPNN) is long and the convergence speed is slow, which leads to a long time for indoor temperature control of buildings based on multi-layer back propagation neural network, resulting in low temperature control efficiency. Summary of the invention

[0004] The purpose of the present invention is to solve the problem of low efficiency in indoor temperature control of existing buildings, and propose a room temperature control method for multi-energy complementary buildings in severe cold regions.

[0005] A method for controlling room temperature of a multi-energy complementary building in a severe cold region comprises the following steps:

[0006] Get the current outdoor temperature and the corresponding indoor temperature, input the outdoor temperature and indoor temperature into the trained NFIN network, and get the result of whether the current indoor temperature is suitable; if the current indoor temperature is too high, lower the indoor temperature; if the current indoor temperature is too low, increase the indoor temperature; if the current indoor temperature is suitable, do not adjust;

[0007] The trained NFIN network is obtained by:

[0008] Step 1: Obtain the building outdoor temperature sequence and the corresponding indoor temperature sequence, set a label for the building outdoor temperature, and use the label, outdoor temperature and corresponding indoor temperature to form a training set;

[0009] Step 2: Use the training set to train the NFIN network to obtain the trained NFIN network.

[0010] Furthermore, the step 1 of setting a label for the outdoor temperature of the building is specifically:

[0011] If the current indoor temperature is less than the first preset temperature threshold, indicating that the indoor temperature is too low, the label of the outdoor temperature data corresponding to the current indoor temperature data is set to 0;

[0012] If the front indoor temperature is greater than or equal to the first preset temperature threshold and less than the second preset temperature threshold, indicating that the temperature is suitable, the outdoor temperature data label corresponding to the current indoor temperature data is set to -1;

[0013] If the current indoor temperature is greater than or equal to the second temperature threshold, indicating that the current temperature is too high, the label of the outdoor temperature data corresponding to the current indoor temperature is set to 1.

[0014] Furthermore, the specific processing process of the NFIN network in step 2 is as follows:

[0015] S1, obtain the membership value of the outdoor temperature to the jth fuzzy set;

[0016] S2, using the membership value of the outdoor temperature belonging to the jth fuzzy set to obtain the matching degree between the outdoor temperature and the fuzzy rule;

[0017] S3, obtain the fuzzy rule number J corresponding to the maximum matching degree between the outdoor temperature and the fuzzy rule, and then compare F J (u) and If the size relationship between Then a new rule is generated, and then S4 is executed; otherwise, the classification result of the indoor temperature corresponding to the outdoor temperature is directly obtained, and then the process ends;

[0018] The classification results of the indoor temperature corresponding to the outdoor temperature include: high, suitable and low;

[0019] Among them, F J (u) is the matching degree between the outdoor temperature sequence and the Jth fuzzy rule, is the preset attenuation threshold;

[0020] S4, updating the Gaussian membership function of the fuzzy rules;

[0021] S5. Defuzzification is performed using the Gaussian membership function of the fuzzy rule obtained in S4 to obtain a classification result of the outdoor temperature corresponding to the indoor temperature.

[0022] Furthermore, the membership degree value of the outdoor temperature obtained in S1 belongs to the jth fuzzy set, specifically:

[0023]

[0024] in, is the i-th outdoor temperature value, yes The membership degree value of j fuzzy sets, m ij is the mean of the jth Gaussian membership function, σ ij is the standard deviation of the jth Gaussian membership function;

[0025] Among them, j=1, 2, and 3 represent the too low temperature fuzzy set, the appropriate temperature fuzzy set, and the too high temperature fuzzy set, respectively.

[0026] Furthermore, the degree of matching between the outdoor temperature and the fuzzy rule is obtained by using the membership value of the outdoor temperature belonging to the jth fuzzy set in S2, specifically:

[0027]

[0028] Among them, F j (u) is the matching degree between the outdoor temperature sequence and the jth fuzzy rule, u is the outdoor temperature sequence, D i =diag(1 / σ i1 ,1 / σ i2 ,....,1 / σ in ) is a diagonal matrix, σ in is the standard deviation of the nth Gaussian membership function, n is the total number of current fuzzy rules, m i =(m i1 ,m i2 ,…,m in ) T is the mean vector of the Gaussian membership function of the i-th outdoor temperature.

[0029] Further, the fuzzy rule number J corresponding to the maximum value of the matching degree between the outdoor temperature and the fuzzy rule is obtained in S3, and then F is compared. J (u) and If the size relationship between Then a new rule is generated, and then S4 is executed; otherwise, the classification result of the indoor temperature corresponding to the outdoor temperature is directly obtained and the process ends, specifically:

[0030] S301, obtaining the fuzzy rule number J corresponding to the maximum value of the matching degree between the outdoor temperature and the fuzzy rule:

[0031]

[0032] S302, Comparison F J (u)≤and The size of Then execute S303; otherwise execute S304;

[0033] Among them, FJ (u) is obtained in the same way as F j (u) is obtained in the same way;

[0034] S303, generate new fuzzy rules, and then execute S4;

[0035] The initial center and initial width of the new fuzzy rule are:

[0036] m (n+1) =u

[0037]

[0038] Among them, m (n+1) is the initial center of the new fuzzy rule, D (n+1) is the initial width of the new fuzzy rule, β≥0, β is the overlap threshold of the Gaussian membership function;

[0039] S304, obtaining the matching degree value F between the current outdoor temperature and the indoor temperature and the fuzzy rule j (u′), then execute S305;

[0040] Get the matching degree value F between the current outdoor temperature and the indoor temperature and the fuzzy rule j (u′), specifically:

[0041]

[0042] Among them, F j (u′) is the matching degree value between the indoor temperature sequence u′ and the jth fuzzy rule, u′ is the indoor temperature sequence, i′ is the indoor temperature data label, D i' =diag(1 / σ i'1 ,1 / σ i'2 ,...,1 / σ i'n ) is a diagonal matrix, σ i'n is the standard deviation of the nth Gaussian membership function, m i' =(m i'1 ,m i'2 ,....,m i'n ) is the mean vector of the Gaussian membership function of the i'th indoor temperature, is the i'th outdoor temperature value, yes The membership degree value of j fuzzy sets, m i'j is the mean of the jth Gaussian membership function, σ i'j is the standard deviation of the jth Gaussian membership function;

[0043] S305: According to the matching degree value F between the outdoor temperature and the fuzzy rule j(u) The matching degree value F between the indoor temperature and the fuzzy rule j (u′), obtain the indoor temperature classification result corresponding to the current outdoor temperature, and end.

[0044] Furthermore, the matching degree value F of the outdoor temperature and the fuzzy rule in S305 is j (u) The matching degree value F between the indoor temperature and the fuzzy rule j (u′), obtain the indoor temperature classification result corresponding to the current outdoor temperature, specifically:

[0045] If F j (u')<0.7 and F j (u)≥0.7, the indoor temperature corresponding to the current outdoor temperature is output to be higher;

[0046] If F j (u')≥0.7 and F j (u)<0.7, then the indoor temperature corresponding to the current outdoor temperature is output to be low;

[0047] If F j (u) and F j If (u′) is less than or equal to 0.7 or greater than or equal to 0.7 at the same time, the indoor temperature corresponding to the current outdoor temperature is output as appropriate.

[0048] Furthermore, the Gaussian membership function of the updated fuzzy rule in S4 is:

[0049]

[0050] Among them, σ i is the standard deviation matrix of the Gaussian membership functions.

[0051] Furthermore, the Gaussian membership function of the fuzzy rule obtained in S4 is used to perform defuzzification in S5 to obtain the classification result of the outdoor temperature corresponding to the indoor temperature, which is specifically:

[0052]

[0053] Among them, y i is the classification result of the indoor temperature corresponding to the current outdoor temperature, a 0i =m 0i is the center of the Gaussian membership function, a ij is the correlation weight coefficient between the i-th outdoor temperature data and the j-th fuzzy rule, u i is the i-th outdoor temperature data.

[0054] Furthermore, the objective function of the NFIN network is as follows:

[0055]

[0056] Among them, y i is the indoor temperature classification result value corresponding to the current outdoor temperature, is the expected indoor temperature classification result value, and RE(j) is the cumulative error of rule j.

[0057] The beneficial effects of the present invention are:

[0058] The present invention proposes a new room temperature control method based on nonlinear finite element network. Based on the influence of outdoor temperature on indoor temperature, the present invention uses the correlation between outdoor temperature and indoor temperature to adjust indoor temperature. The premise part of the rule of the present invention adopts input clustering method, and the post-part part adopts flexible linear combination structure, which reduces the number of rules and the number of parameters in each rule. The parameter adjustment of the NFIN network proposed by the present invention conforms to the temperature changes in different environments, has higher sensitivity to temperature changes, can shorten the temperature control time, and improves the control efficiency of indoor temperature. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is the NFIN network structure diagram;

[0060] Figure 2 It is the schematic diagram of fuzzy controller;

[0061] Figure 3 This is the hardware framework diagram. DETAILED DESCRIPTION

[0062] Specific implementation method 1: This implementation method is a method for controlling room temperature of a multi-energy complementary building in a severe cold region. The specific process is: obtain the current outdoor temperature and the corresponding indoor temperature, input the outdoor temperature and the indoor temperature into the trained NFIN network, and obtain the result of whether the current indoor temperature is suitable; if the current indoor temperature is too high, reduce the indoor temperature; if the current indoor temperature is too low, increase the indoor temperature; if the current indoor temperature is suitable, do not adjust;

[0063] The trained NFIN network is obtained by:

[0064] Step 1: Obtain the building outdoor temperature sequence and the corresponding indoor temperature sequence, set a label for the building outdoor temperature, and use the label, outdoor temperature and corresponding indoor temperature to form a training set;

[0065] The outdoor temperature label indicates the classification result of the indoor temperature corresponding to the current outdoor temperature, including: -1, 0, 1; 0 means low, -1 means suitable, 1 means high;

[0066] Set a label for the outdoor temperature data according to the current outdoor corresponding indoor temperature data, specifically:

[0067] If the current indoor temperature is less than the first preset temperature threshold, indicating that the indoor temperature is too low, the outdoor temperature data corresponding to the current indoor temperature data is set to a too low label, that is, set to 0;

[0068] If the front indoor temperature is greater than or equal to the first preset temperature threshold and less than the second preset temperature threshold, it means the temperature is suitable, then the outdoor temperature data corresponding to the current indoor temperature data is set with a suitable label, that is, set to -1;

[0069] If the current indoor temperature is greater than or equal to the second temperature threshold, it means that the current temperature is too high, and the outdoor temperature data corresponding to the current indoor temperature is set with an overly high label, that is, set to 1;

[0070] Step 2: Use the training set to train the NFIN network to obtain the trained NFIN network;

[0071] The specific processing process of the NFIN network is as follows:

[0072] S1. Obtain the membership value of the outdoor temperature to the jth fuzzy set, specifically:

[0073]

[0074] in, is the i-th outdoor temperature value, yes The membership degree value of j fuzzy sets, m ij is the mean of the jth Gaussian membership function, σ ij is the standard deviation of the jth Gaussian membership function;

[0075] Among them, j = 1, 2, 3 represent the too low temperature fuzzy set, the appropriate temperature fuzzy set and the too high temperature fuzzy set respectively;

[0076] The low temperature fuzzy set stores outdoor temperature data with the label of too low temperature, the suitable temperature fuzzy set stores outdoor temperature data with the label of suitable temperature, and the too high temperature fuzzy set stores outdoor temperature data with too low temperature; unlike other clustering-based partitioning methods, in which each input variable has the same number of fuzzy sets, the number of fuzzy sets for each input variable is not necessarily the same in NFIN.

[0077] S2. Using the membership value of the outdoor temperature belonging to the jth fuzzy set, the matching degree between the outdoor temperature and the fuzzy rule is obtained, that is, the membership function of the fuzzy rule, which is specifically:

[0078]

[0079] Among them, F j(u) is the matching degree between the outdoor temperature sequence and the jth fuzzy rule, u is the outdoor temperature sequence, D i =diag(1 / σ i1 ,1 / σ i2 ,....,1 / σ in ) is a diagonal matrix, σ in is the standard deviation of the nth Gaussian membership function, n is the total number of current fuzzy rules, m i =(m i1 ,m i2 ,....,m in ) T is the mean vector of the Gaussian membership function of the i-th outdoor temperature;

[0080] F j The value of (u) is between 0 and 1;

[0081] S3, obtain the fuzzy rule number J corresponding to the maximum value of the matching degree between the outdoor temperature and the fuzzy rule, if Then generate a new rule and then execute S4; otherwise, match the current indoor temperature with the existing fuzzy rules to obtain the classification result of the indoor temperature data corresponding to the current outdoor temperature, and end; specifically:

[0082] S301, obtaining the fuzzy rule number J corresponding to the maximum value of the matching degree between the outdoor temperature and the fuzzy rule:

[0083]

[0084] Where n is the total number of current fuzzy rules;

[0085] S302, Comparison F J and The size of Then execute S303; otherwise execute S304;

[0086] Among them, F J (u) is obtained in the same way as F in S2 j (u)Same.

[0087] in, is a pre-specified threshold value that decays during the learning process, i.e., a preset decay threshold;

[0088] S303, generate new fuzzy rules, and then execute S4, specifically:

[0089] The initial center and initial width of the new fuzzy rule are:

[0090] m (n+1) =u

[0091]

[0092] Among them, m (n+1) is the initial center of the new fuzzy rule, D (n+1) is the initial width of the new fuzzy rule, β≥0 is the threshold of the overlap degree of the Gaussian membership function;

[0093] The fuzzy rule form is:

[0094] Rulej:=IFu1isinA i1 and…andu N isinA in

[0095] THEN i iV i

[0096] Among them, u1 is the first outdoor temperature data, A in is the nth fuzzy set, j is the label of the fuzzy set, u n is the Nth outdoor temperature, y i is the classification result value (0, -1, 1) of the corresponding indoor temperature of the current outdoor temperature data, V i is the output variable y i The position of the symmetric membership function, whose width is ignored in the defuzzification process;

[0097] S304, obtaining the matching degree value between the current outdoor temperature and the indoor temperature and the fuzzy rule, and then executing S305, specifically:

[0098]

[0099] Among them, F j (u′) is the matching degree value between the indoor temperature sequence u′ and the jth fuzzy rule, u′ is the indoor temperature sequence, i′ is the indoor temperature data label, D i' =diag(1 / σ i'1 ,1 / σ i'2 ,...,1 / σ i'n ) is a diagonal matrix, σ i'n is the standard deviation of the nth Gaussian membership function, n is the number of current fuzzy rules, m i' =(m i'1 ,m i'2 ,....,m i'n ) is the mean vector of the Gaussian membership function of the i'th indoor temperature, is the i'th outdoor temperature value, yes The membership degree value of j fuzzy sets, m i'j is the mean of the jth Gaussian membership function, σ i'jis the standard deviation of the jth Gaussian membership function;

[0100] S305: According to the matching degree value F between the outdoor temperature and the fuzzy rule j (u) The matching degree value F between the indoor temperature and the fuzzy rule j (u′), obtain the indoor temperature classification result corresponding to the current outdoor temperature, and end directly:

[0101] If F j (u')<0.7 and F j (u)≥0.7, the indoor temperature corresponding to the current outdoor temperature is output to be higher;

[0102] If F j (u')≥0.7 and F j (u)<0.7, then the indoor temperature corresponding to the current outdoor temperature is output to be low;

[0103] If F j (u) and F j If (u′) is less than or equal to 0.7 or greater than or equal to 0.7 at the same time, the indoor temperature corresponding to the current outdoor temperature is output as appropriate.

[0104] S4. Update the Gaussian membership function of the fuzzy rule:

[0105]

[0106] Among them, σ i is the standard deviation matrix of the Gaussian membership function;

[0107] S5. Defuzzification is performed using the Gaussian membership function of the fuzzy rule obtained in S4 to obtain a defuzzification result, i.e., a classification result of the outdoor temperature corresponding to the indoor temperature:

[0108]

[0109] Among them, y i is the classification result of the indoor temperature corresponding to the current outdoor temperature, a 0i =m 0i is the center of the Gaussian membership function, a ij is the correlation weight coefficient between the i-th outdoor temperature data and the j-th fuzzy rule, u i is the i-th outdoor temperature data;

[0110] In this embodiment, two types of nodes are used in S5, and they are Figure 1They are represented as blank and shaded circles, respectively, for functional distinction as described below. The nodes represented by blank circles (blank nodes) are the basic nodes representing the fuzzy sets (described by Gaussian membership functions) of the output variables. Since the widths of the Gaussian membership functions are equal, only the center of each Gaussian membership function is passed to the next layer for the local maximum mean (LMOM) defuzzification operation. The function of width is only used for output clustering. Different nodes in layer 3 can be connected to the same blank node in layer 4, which means that the same resulting fuzzy set is specified for different rules. The role of this fuzzy set is that the user transforms the membership of different nodes in order to classify them into the same set. The formula for the blank node is:

[0111]

[0112] where a 0i =m 0i , the center of the Gaussian membership function. As for the shadow node, it is generated only when necessary. Each node in layer 3 has its corresponding colored node in layer 4. One input of the shadow node is the output passed from layer 3, while the other possible inputs (terms) are input variables from layer 1, such as Figure 1 As shown. The shadow node function formula is:

[0113]

[0114] Among them, the sum is the sum of all inputs. By combining these two types of nodes, the present invention obtains the entire function performed by this layer for each rule. The formula is as follows:

[0115]

[0116] From the above equation, by creating shadow nodes, the consequent part performs the same function as TSK type fuzzy rules, where the consequent is a linear combination of the input variables. Therefore, we connect the inputs to the shadow nodes in NFIN.

[0117] Specific implementation method 2: NFIN network adopts the following objective function:

[0118]

[0119] Among them, y i is the indoor temperature classification result value corresponding to the current outdoor temperature, is the expected indoor temperature classification result value, RE(j) is the cumulative error of rule j;

[0120] Embodiment: In order to verify the beneficial effects of the present invention, the present invention carried out the following experiments:

[0121] like Figure 2As shown in the figure, the indoor automatic temperature control device consists of three parts: personal computer (PC), temperature control board, and temperature sensor. The hardware includes: signal sampling circuit, single chip basic system (8031), A / D conversion circuit, keyboard and display circuit, and execution circuit to build a room temperature controller. Figure 3 As shown. This embodiment uses a PC equipped with an Intel Pentium II 600 processor as the control core, and runs the control program to realize the control and data processing functions of the entire system. A PC equipped with an Intel Pentium I11600 processor is used in the room temperature controller. The control program is written in Borland C language, and the output voltage is controlled by the temperature control board to be sent to the control switch. In order to realize the temperature measurement and the sending of control signals, a temperature control card based on the PCI-1710 card is used. The PCI-1710 card is developed by Advantech. It is a multi-function data acquisition card with a PCI bus interface, providing multiple 12-bit AD and D / A channels. For temperature measurement, the PTIOO sensor is connected to the temperature control board to measure indoor and outdoor temperatures. If u(k)<5 voltage, then P(k)=0% the heater is off, if u(k)<2 voltage, then P(k)=100% the regulator is on, if 0

Claims

1. A method for controlling room temperature of a multi-energy complementary building in severe cold regions, characterized in that The specific process of the method is: Get the current outdoor temperature and the corresponding indoor temperature, input the outdoor temperature and indoor temperature into the trained NFIN network, and get the result of whether the current indoor temperature is suitable; If the current indoor temperature is too high, lower the indoor temperature; if the current indoor temperature is too low, raise the indoor temperature; if the current indoor temperature is appropriate, do not adjust; The trained NFIN network is obtained by: Step 1: Obtain the building outdoor temperature sequence and the corresponding indoor temperature sequence, set a label for the building outdoor temperature, and use the label, outdoor temperature and corresponding indoor temperature to form a training set; Step 2: Use the training set to train the NFIN network to obtain the trained NFIN network.

2. The method for controlling room temperature of a multi-energy complementary building in severe cold regions according to claim 1, characterized in that: The step 1 is to set a label for the outdoor temperature of the building, specifically: If the current indoor temperature is less than the first preset temperature threshold, indicating that the indoor temperature is too low, the label of the outdoor temperature data corresponding to the current indoor temperature data is set to 0; If the front indoor temperature is greater than or equal to the first preset temperature threshold and less than the second preset temperature threshold, indicating that the temperature is suitable, the outdoor temperature data label corresponding to the current indoor temperature data is set to -1; If the current indoor temperature is greater than or equal to the second temperature threshold, indicating that the current temperature is too high, the label of the outdoor temperature data corresponding to the current indoor temperature is set to 1.

3. The room temperature control method of a multi-energy complementary building in severe cold regions according to claim 2 is characterized in that: The specific processing process of the NFIN network in step 2 is as follows: S1, obtain the membership value of the outdoor temperature to the jth fuzzy set; S2, using the membership value of the outdoor temperature belonging to the jth fuzzy set to obtain the matching degree between the outdoor temperature and the fuzzy rule; S3, obtain the fuzzy rule number J corresponding to the maximum matching degree between the outdoor temperature and the fuzzy rule, and then compare F J (u) and If the size relationship between Then generate new rules and then execute S4; Otherwise, directly obtain the classification result of the indoor temperature corresponding to the outdoor temperature and end; The classification results of the indoor temperature corresponding to the outdoor temperature include: high, suitable and low; Among them, F J (u) is the matching degree between the outdoor temperature sequence and the Jth fuzzy rule, is the preset attenuation threshold; S4, updating the Gaussian membership function of the fuzzy rules; S5. Defuzzification is performed using the Gaussian membership function of the fuzzy rule obtained in S4 to obtain a classification result of the outdoor temperature corresponding to the indoor temperature.

4. The method for controlling room temperature of a multi-energy complementary building in severe cold regions according to claim 3 is characterized in that: The membership degree value of the outdoor temperature obtained in S1 belongs to the jth fuzzy set, specifically: in, is the i-th outdoor temperature value, yes The membership degree value of j fuzzy sets, m ij is the mean of the jth Gaussian membership function, σ ij is the standard deviation of the jth Gaussian membership function; Among them, j=1, 2, and 3 represent the too low temperature fuzzy set, the appropriate temperature fuzzy set, and the too high temperature fuzzy set, respectively.

5. The method for controlling room temperature of a multi-energy complementary building in severe cold regions according to claim 4 is characterized in that: The degree of matching between the outdoor temperature and the fuzzy rule is obtained by using the membership value of the outdoor temperature belonging to the jth fuzzy set in S2, specifically: Among them, F j (u) is the matching degree between the outdoor temperature sequence and the jth fuzzy rule, u is the outdoor temperature sequence, D i =diag(1 / σ i1 ,1 / σ i2 ,…,1 / σ in ) is a diagonal matrix, σ in is the standard deviation of the nth Gaussian membership function, n is the total number of current fuzzy rules, m i =(m i1 ,m i2 ,…,m in ) T is the mean vector of the Gaussian membership function of the i-th outdoor temperature.

6. A method for controlling room temperature of a multi-energy complementary building in severe cold regions according to claim 5, characterized in that: The fuzzy rule number J corresponding to the maximum value of the matching degree between the outdoor temperature and the fuzzy rule is obtained in S3, and then F is compared. J (u) and If the size relationship between Then a new rule is generated, and then S4 is executed; otherwise, the classification result of the indoor temperature corresponding to the outdoor temperature is directly obtained and the process ends, specifically: S301, obtaining the fuzzy rule number J corresponding to the maximum value of the matching degree between the outdoor temperature and the fuzzy rule: S302, Comparison F J (u)≤and The size of Then execute S303; otherwise execute S304; Among them, F J (u) is obtained in the same way as F j (u) is obtained in the same way; S303, generate new fuzzy rules, and then execute S4; The initial center and initial width of the new fuzzy rule are: m (n+1) =u Among them, m (n+1) is the initial center of the new fuzzy rule, D (n+1) is the initial width of the new fuzzy rule, β≥0, β is the overlap threshold of the Gaussian membership function; S304, obtaining the matching degree value F between the current outdoor temperature and the indoor temperature and the fuzzy rule j (u′), then execute S305; Get the matching degree value F between the current outdoor temperature and the indoor temperature and the fuzzy rule j (u′), specifically: Among them, F j (u′) is the matching degree value between the indoor temperature sequence u′ and the jth fuzzy rule, u′ is the indoor temperature sequence, i′ is the indoor temperature data label, D i' =diag(1 / σ i'1 ,1 / σ i'2 ,...,1 / σ i'n ) is a diagonal matrix, σ i'n is the standard deviation of the nth Gaussian membership function, m i' =(m i'1 ,m i'2 ,....,m i'n ) is the mean vector of the Gaussian membership function of the i'th indoor temperature, is the i'th outdoor temperature value, yes The membership degree value of j fuzzy sets, m i'j is the mean of the jth Gaussian membership function, σ i'j is the standard deviation of the jth Gaussian membership function; S305: According to the matching degree value F between the outdoor temperature and the fuzzy rule j (u) The matching degree value F between the indoor temperature and the fuzzy rule j (u′), obtain the indoor temperature classification result corresponding to the current outdoor temperature, and end.

7. A method for controlling room temperature of a multi-energy complementary building in severe cold regions according to claim 6, characterized in that: The matching degree value F of the outdoor temperature and the fuzzy rule in S305 j (u) The matching degree value F between the indoor temperature and the fuzzy rule j (u′), obtain the indoor temperature classification result corresponding to the current outdoor temperature, specifically: If F j (u')<0.7 and F j (u)≥0.7, the indoor temperature corresponding to the current outdoor temperature is output to be higher; If F j (u')≥0.7 and F j (u)<0.7, then the indoor temperature corresponding to the current outdoor temperature is output to be low; If and F j (u) and F j If (u′) is less than or equal to 0.7 or greater than or equal to 0.7 at the same time, the indoor temperature corresponding to the current outdoor temperature is output as appropriate.

8. The method for controlling room temperature of a multi-energy complementary building in severe cold regions according to claim 7, characterized in that: The Gaussian membership function of the updated fuzzy rule in S4 is: Among them, σ i is the standard deviation matrix of the Gaussian membership functions.

9. A method for controlling room temperature of a multi-energy complementary building in severe cold regions according to claim 8, characterized in that: In S5, the Gaussian membership function of the fuzzy rule obtained in S4 is used to perform defuzzification to obtain the classification result of the outdoor temperature corresponding to the indoor temperature, which is specifically: Among them, y i is the classification result of the indoor temperature corresponding to the current outdoor temperature, a 0i =m 0i is the center of the Gaussian membership function, a ij is the correlation weight coefficient between the i-th outdoor temperature data and the j-th fuzzy rule, u i is the i-th outdoor temperature data.

10. A method for controlling room temperature of a multi-energy complementary building in severe cold regions according to claim 9, characterized in that: The objective function of the NFIN network is as follows: Among them, y i is the indoor temperature classification result value corresponding to the current outdoor temperature, is the expected indoor temperature classification result value, and RE(j) is the cumulative error of rule j.

Images