A markov-based parameter identification method for internal energy consumption model of underground building

Abstract

The application provides a Markov-based underground building internal energy consumption model parameter identification method, and specifically as follows: energy consumption model forms of an air conditioner dehumidifier unit, a circulating water pump and a cooling tower fan in an underground building are respectively established, forms of Markov models including dimensions of a state vector, dimensions of an input vector, a state transition matrix and an input matrix are determined, data required for model parameter identification is collected, for given data and models, least square method is used to estimate model parameters, and root mean square error and absolute percentage error are used to evaluate identification results, so that a set of energy consumption models suitable for underground building internal equipment is obtained. The application can solve the problem that the air conditioner dehumidifier unit, the circulating water pump and the fan cannot directly use rated parameters for energy consumption calculation due to complex operation characteristics and operation conditions in the prior art, and lays a good foundation for later underground building internal equipment energy consumption calculation and operation optimization.

Description

Technical Field

[0001] The present invention relates to an energy consumption model of equipment inside an underground building, in particular to a parameter identification method of an energy consumption model inside an underground building based on Markov. Background Art

[0002] According to 2019 statistics from the International Energy Agency, the construction industry, one of the three largest energy users, accounts for 38% of global energy consumption and 39% of global carbon dioxide emissions. my country has nearly 40 billion square meters of existing buildings, of which only 5% are energy-efficient. Nearly 2 billion square meters are newly built annually, but less than 100 million square meters are truly energy-efficient. Furthermore, with the rapid development of modern cities in my country, urbanization is occurring at an unprecedented pace and scale, exacerbating the shortage of surface space resources. To alleviate this problem, the development and utilization of urban underground space has gradually become widespread over the past 20 years. Underground space, which can provide 25% to 40% additional space, is gaining increasing attention, and semi-basement areas in residential buildings are also being widely developed and utilized.

[0003] Due to the lack of natural light and ventilation, underground buildings require long-term use of equipment such as air conditioning and dehumidification units, water pumps, and fan systems to meet daily needs. In order to better achieve optimized equipment management and energy conservation, the use of energy consumption models for analysis and prediction is a more scientific technical means. Existing technologies directly use rated parameters to calculate energy consumption and only use historical energy consumption data to establish energy consumption models for prediction. However, due to the complex operating characteristics and operating conditions of air conditioning and dehumidification units, water pumps, and fan systems, there are differences between the actual characteristics of each device and the rated characteristics at the time of leaving the factory. The results of directly calculating energy consumption using rated parameters cannot reflect the actual operating energy consumption. The method of using only historical energy consumption data to establish an energy consumption model for prediction has a certain accuracy rate, but it does not link energy consumption with the characteristics and parameters of each device, which reduces the accuracy of the prediction results. Summary of the Invention

[0004] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a Markov-based method for identifying parameters of energy consumption models inside underground buildings, so as to solve the problem in the prior art that air-conditioning and dehumidification units, circulating water pumps and fans inside underground buildings cannot be directly calculated using rated parameters due to the complex operating characteristics and operating conditions. This method collects real data from the system and uses the measured data of equipment energy consumption to establish an energy consumption model for each device, identifies the energy consumption model parameters based on Markov, and verifies and simulates the identified model in combination with the measured data. It can accurately predict energy consumption and establish a mapping relationship between equipment parameters and energy consumption, which facilitates energy consumption analysis and optimization management, provides a scientific basis for better realizing energy consumption model optimization and equipment energy-saving control, lays a solid foundation, and also plays a positive role in the maintenance and expansion of the later system.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] A Markov-based parameter identification method for an underground building internal energy consumption model comprises the following steps:

[0007] S1. Based on the operating conditions of the air conditioning and dehumidification units, circulating water pumps, and cooling tower fans inside the underground building and the operating characteristics of each device, energy consumption models for the air conditioning and dehumidification units, circulating water pumps, and cooling tower fans are established respectively;

[0008] S2. Determine the form of the Markov model corresponding to each device, including the dimension of the state vector, the dimension of the input vector, the shape of the state transition matrix, and the input matrix;

[0009] S3. Collect input data required for model parameter identification of each device, where the input data includes a state sequence of each set of state vectors and a corresponding observation value sequence;

[0010] S4. Parameter estimation: For a given input data and model, the least squares method is used to estimate the model parameters and calculate the parameters to be identified for each device model;

[0011] S5. Model verification and evaluation: the method is as follows: input the identified model parameters into the model in step S1 for calculation, draw a comparison chart between the calculation results and the actual measurement results, and then evaluate the performance of the model.

[0012] In step S1, the energy consumption of the air conditioning and dehumidification unit is mainly affected by the cooling water inlet temperature Twi and the outlet temperature T wo , dehumidifier unit humidity RH and air supply temperature T ai and return air temperature T ao Therefore, the energy consumption model of the dehumidifier unit is expressed as follows

[0013] P=f(RH,T wo , T wi , T ao , T ai ) (1)

[0014] In order to facilitate operation, the difference between the inlet and outlet temperatures of the cooling water ΔT is used. w The difference between the supply and return air temperature ΔT a calculate,

[0015] ΔT w =T wo -T wi (2)

[0016] ΔT a =T ai -T ao (3)

[0017] Formulas (1), (2), and (3) are fitted into the following quadratic polynomial:

[0018] P=x1+x2*ΔT w +x3*RH+x4*ΔT a +x5*ΔT w 2 +x6*RH 2 +x7*ΔT a 2 +x8*ΔT w *RH+x9*ΔT w *ΔT a +x 10 *RH*ΔT a (4)

[0019] In the above formula, x1~x 10 are the parameters to be identified for the air conditioning and dehumidification unit;

[0020] In step S1, the circulating water pump is similar to the chilled water pump. According to the characteristic relationship between the head and flow of the circulating water pump at non-rated speed,

[0021]

[0022]

[0023] In the above formula, c0~c2, d0~d2 are the parameters to be identified of the circulating water pump;

[0024] Therefore, the mathematical expression of the circulating water pump energy consumption model is:

[0025]

[0026] In the above formula, m cwis the current circulating water pump flow, H cw is the current circulating water pump head, η cw is the operating point efficiency of the circulating water pump, g d is the flow head coefficient; n is the operating point speed, n nom is the rated speed;

[0027] In step S1, the wind pressure and efficiency of the cooling tower fan can be approximately expressed as a quadratic function related to the air volume:

[0028]

[0029]

[0030] In the above formula, F is the wind pressure of the cooling tower fan; η is the efficiency of the cooling tower fan; Q f is the cooling tower fan air volume, m3 / h; l1, l2, l3, l4, l5, l6 are the parameters to be identified of the cooling tower fan, so the power consumption P of the cooling tower fan is expressed as the following function:

[0031] P=Q f *F / (3600*1000*η) (12)

[0032] In step S2, the dimension of the state vector, the dimension of the input vector, the shape of the state transfer matrix and the input matrix are selected according to the model properties and actual operating conditions; specifically, in the energy consumption model of the air-conditioning and dehumidification unit, the number of parameters to be identified in the model is 10, then the dimension of the state vector and the dimension of the input vector in the corresponding Markov model are both 10; the state transfer matrix is ​​in the form of a 10*10 matrix, and the input matrix is ​​in the form of a 10*1 matrix; the number of parameters to be identified in the circulating water pump model is 6, so the dimension of the state vector and the dimension of the input vector of the circulating water pump model are both 6, the state transfer matrix is ​​in the form of a 6*6 matrix, and the input matrix is ​​in the form of a 6*1 matrix; the number of parameters to be identified in the cooling tower fan model is 6, so the dimension of the state vector and the dimension of the input vector of the cooling tower fan model are both 6, the state transfer matrix is ​​in the form of a 6*6 matrix, and the input matrix is ​​in the form of a 6*1 matrix.

[0033] In step S3, in the energy consumption model of the air-conditioning and dehumidification unit, the state sequence and the corresponding observation value sequence are determined based on the historical operating energy consumption data of the dehumidification unit; in the circulating water pump model, the state sequence and the corresponding observation value sequence are determined based on the historical operating energy consumption data of the circulating water pump; in the cooling tower fan model, the state sequence and the corresponding observation value sequence are determined based on the historical operating energy consumption data of the cooling tower fan.

[0034] In step S4, the least square method is used to estimate the parameters of the given data and the Markov model, and the parameters to be identified in the Markov model are obtained by minimizing the residual sum of squares.

[0035] In step S5, the root mean square error and absolute percentage error are used as evaluation indicators to compare and analyze the model operation results with the actual operation energy consumption data to evaluate the accuracy and error of the model after parameter identification. Compared with other time series prediction evaluation indicators, MASE has the following advantages:

[0036] ① The prediction accuracy between different time series can be compared because MASE takes into account the fluctuation range of historical observations.

[0037] ②MASE has less impact on outliers because MASE uses absolute error instead of square error.

[0038] When the results of both evaluation indicators are close to 0, it can be determined that the equipment model has high accuracy, small error, and relatively accurate parameter identification results.

[0039] In summary, the present invention can establish a mechanism model by using the actual operating data of the air-conditioning and dehumidification units, circulating water pumps and fan equipment collected, using the equipment operating characteristics, performing parameter identification based on the Markov model, and evaluating the model identification results, thereby solving the problem in the prior art that the air-conditioning and dehumidification units, circulating water pumps and fans cannot directly use rated parameters for energy consumption calculation due to the complex operating characteristics and operating conditions.

[0040] Compared with the prior art, the present invention has at least the following beneficial effects:

[0041] The present invention is a Markov-based method for identifying parameters of an underground building internal energy consumption model. A Markov model is a mathematical model that describes a random process. It has the Markov property, meaning that the future state depends only on the current state and is independent of the historical state. It has the following advantages:

[0042] ① Non-parametric: Markov parameter identification does not require any prior assumptions about the system's linearity, nonlinearity, time-invariance or time-varying nature, and is therefore highly adaptable and flexible.

[0043] ② Simple model structure: Models established by Markov parameter identification are usually simple state space models, so the model structure is simple and easy to implement.

[0044] ③High real-time performance: The Markov parameter identification method can identify and predict the system in real time, and has good adaptability to certain control systems that require rapid response.

[0045] ④ No need to impose strict precondition constraints on input and output signals: Markov parameter identification does not require special constraints on input and output signals. The Markov parameter identification method can be applied to any random process as long as certain basic conditions are met.

[0046] ⑤Can process non-stationary signals: Markov parameter identification can process non-stationary signals and thus has a wider range of applications, such as being widely used in control systems, signal processing and communications.

[0047] ⑥ This method can identify relevant parameters that the manufacturer fails to provide through experimental data. According to historical operating data and equipment operating characteristics, the Markov model and least squares theory are used to identify the parameters of the air-conditioning and dehumidification unit model, circulating water pump model and fan model in the internal operating equipment of underground buildings, and obtain a more accurate mathematical model. It provides an accurate match with the actual performance under a wide range of operating conditions. It is simple and easy to implement in actual projects, and can ensure both optimization accuracy and model simplicity, laying a good foundation for the subsequent energy consumption calculation and operation optimization of internal equipment in underground buildings. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 Flowchart for parameter identification of energy consumption model for equipment inside underground buildings;

[0049] Figure 2 This is a comparison chart between the actual value and the calculated value of the air conditioning and dehumidification unit;

[0050] Figure 3 This is a comparison chart of the actual and calculated values ​​of the pump flow-head;

[0051] Figure 4 This is a comparison chart of the actual and calculated values ​​of the air volume-pressure of fan 1;

[0052] Figure 5 This is a comparison chart of the actual and calculated values ​​of the air volume-pressure of fan 2. DETAILED DESCRIPTION

[0053] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments.

[0054] The present invention provides a Markov-based parameter identification method for an underground building internal energy consumption model. The Markov model is used to perform parameter identification on the parameters to be identified in the air-conditioning dehumidification unit, circulating water pump and fan models. First, the energy consumption mechanism models of the dehumidification unit, water pump and fan are respectively established according to the operating principles of the equipment in the underground building; then the Markov model is established, and the dimension of the state vector, the dimension of the input vector, the shape of the state transfer matrix and the input matrix in the model need to be determined according to the model characteristics and actual operating conditions; then the collected equipment operation data is divided into a state sequence and a corresponding observation value sequence for subsequent parameter identification and model evaluation; then the least squares method is used to perform parameter identification on the parameters in the given Markov model; finally, the model accuracy of the identified model is verified using the root mean square error and the absolute percentage error as evaluation indicators.

[0055] See also Figure 1 The present invention provides a Markov-based method for identifying parameters of an underground building internal energy consumption model, comprising the following steps:

[0056] S1. Based on the operation of the energy consumption model inside the underground building and the operating characteristics of each device, establish energy consumption models for the air conditioning and dehumidification unit, circulating water pump, and cooling tower fan respectively;

[0057] The energy consumption of air conditioning and dehumidification units is mainly affected by the cooling water inlet temperature (T wi ) and outlet water temperature (T wo ), dehumidifier unit humidity (RH) and air supply temperature (T ai ) and return air temperature (T ao ), the energy consumption model of the dehumidifier can be expressed as follows:

[0058] P=f(RH,T wo , T wi , T ao， T ai ) (1)

[0059] In order to facilitate operation, the difference in cooling water inlet and outlet temperature (ΔT w ) and the difference between the supply and return air temperatures (ΔT a )calculate.

[0060] ΔT w =Tw o -Tw i (2)

[0061] ΔT a =T ai -T ao (3)

[0062] Formulas (1), (2), and (3) can be fitted into the following quadratic polynomials:

[0063] P=x1+x2*ΔT w +x3*RH+x4*ΔT a +x5*ΔT w 2 +x6*RH 2 +x7*ΔT a 2 +x8*ΔT w *RH+x9*ΔT w *ΔT a +x 10 *RH*ΔT a (4)

[0064] In the above formula, x1~x 10 are the parameters to be identified for the air conditioning and dehumidification unit.

[0065] Also consider the energy consumption of the circulating water pump, similar to the chilled water pump. When running at rated speed, the flow-head and flow-efficiency relationship of the cooling water pump in the cooling water system is determined by the following formula

[0066]

[0067]

[0068] In the above formula, m cw，nom is the rated flow rate of the pump, kg / s, H cw，nom is the rated head of the water pump, m; similarly, according to the characteristic relationship between the water pump head and flow at non-rated speed, the characteristic relationship between the circulating water pump head and flow at non-rated speed is:

[0069]

[0070]

[0071] In the above formula, c0~c2, d0~d2 are the parameters to be identified of the circulating water pump; m cw is the current circulating water pump flow, H cw is the current circulating water pump head, η cw is the working point efficiency of the circulating water pump, n is the working point speed, n nom is the rated speed;

[0072] Therefore, the mathematical expression of the circulating water pump energy consumption model is:

[0073]

[0074] In the above formula, g d is the flow head coefficient.

[0075] The wind pressure and efficiency of the cooling tower fan can be approximately expressed as a quadratic function related to the air volume:

[0076]

[0077]

[0078] Where, F is the wind pressure of the cooling tower fan; η is the efficiency of the cooling tower fan; Q f is the cooling tower fan air volume, m 3 / h; l1, l2, l3, l4, l5, l6 are the cooling tower fan standby

[0079] Identification parameters. Therefore, the power consumption of the cooling tower fan can be regarded as the following function:

[0080] P=Q f *F / (3600*1000*η) (12)

[0081] S2. Determine the form of the Markov model, including the dimensions of the state vector, the dimensions of the input vector, the shape of the state transfer matrix, and the input matrix. The vector dimensions and matrix shape need to be selected based on the nature of the model and the actual operation. The result of the state transfer matrix of the Markov model at time n is only related to the result at time n-1. Therefore, the n-time Markov model can be regarded as the following function:

[0082]

[0083] Where z is the number of parameters to be identified in the energy consumption model of underground building internal equipment, a n，z is the calculation result of parameter z at the nth moment. In the energy consumption model of the air-conditioning dehumidification unit, as shown in formula (4), the number of parameters to be identified in the model is 10, so the dimension of the state vector and the dimension of the input vector in the corresponding Markov model are both 10; the state transfer matrix is ​​in the form of a 10*10 matrix, and the input matrix is ​​in the form of a 10*1 matrix. The observation value sequence is determined according to the historical operation energy consumption data of the dehumidification unit in Table 1; as shown in formulas (7) to (9), the number of parameters to be identified in the water pump model is 6, so the state vector of the water pump model is of dimension The dimensions of the state vector and the input vector are both 6, the state transfer matrix is ​​in the form of a 6*6 matrix, the input matrix is ​​in the form of a 6*1 matrix, and the observation value sequence is determined according to the historical operation energy consumption data of the dehumidifier unit in Table 2; as shown in Equations (10) to (12), the number of parameters to be identified in the fan model is 6, so the dimensions of the state vector of the model and the input vector are both 6, the state transfer matrix is ​​in the form of a 6*6 matrix, the input matrix is ​​in the form of a 6*1 matrix, and the observation value sequence is determined according to the historical operation energy consumption data of the dehumidifier unit in Table 3.

[0084] S3. Collect the data required for model parameter identification. The data includes the system state sequence and the corresponding observation value sequence. The number of parameters to be identified for the dehumidifier model is 10, and the number of parameters to be identified for the water pump model and fan model is 6. The observation value sequence is the historical energy consumption data of the underground building. 10 are the parameters to be identified for the air conditioning and dehumidification unit. c0~c2, d0~d2 in the circulating water pump model are the parameters to be identified for the circulating water pump. l1~l6 are the parameters to be identified for the fan model.

[0085] Some input data of the equipment are shown in Tables 1 to 3.

[0086] Table 1 Input data of air conditioning and dehumidification unit

[0087]

[0088] Table 2 Water pump input data

[0089]

[0090]

[0091] Table 3 Fan input data

[0092]

[0093] S4. For given data and model, use the least squares method to estimate the model parameters.

[0094] The least squares method provides the best fitting method for experimental data in the sense of minimum variance for simple linear system problems. It is a common mathematical method. Assume that the input data of the system obtained by sampling is an n-dimensional variable The output observation value is an m-dimensional variable

[0095] Y=(y1 y2 … y m ). The linear relationship between the two is as follows:

[0096] Y=φθ (14)

[0097] Where, θ=(θ1 θ2 … θ n ) is a set of unknown constant parameters. In order to make the fitting model close to the actual load, the error vector is defined Set the objective function:

[0098]

[0099] When the objective function J is minimum,

[0100]

[0101] Calculation results It is the least squares estimation of the parameter θ.

[0102] After parameter identification, the identification results of the equipment model are shown in Table 4-6.

[0103] Table 4 Parameter identification results of air conditioning and dehumidification units

[0104]

[0105] Table 5 Pump parameter identification results

[0106]

[0107] Table 6 Fan parameter identification results

[0108]

[0109] According to the identification results, the energy consumption model of the air conditioning and dehumidification unit is expressed as:

[0110] P=4.502-0.884*ΔT w +0.678*RH+4.517*ΔT a -0.039*ΔT w 2

[0111] -0.007*RH 2 -0.255*ΔT a 2 +0.035*ΔT w *RH+0.069*ΔT w *ΔT a

[0112] -0.079*RH*ΔT a (17)

[0113] The energy consumption model of the water pump is expressed as:

[0114]

[0115]

[0116]

[0117] The energy consumption model of fans 1 and 2 is expressed as:

[0118]

[0119]

[0120] P1=Qf *F1 / (3600*1000*η) (23)

[0121]

[0122]

[0123] P2=Q f *F2 / (3600*1000*η) (26)

[0124] S5. Model validation, evaluation indicators include root mean square error and absolute percentage error;

[0125] The identified model was compared and analyzed, and the calculated results were plotted against the actual measured results to evaluate the model's performance. The evaluation metrics were root mean square error (RMSE) and absolute percentage error. The root mean square error (RMSE) is a commonly used metric for measuring model prediction accuracy. It measures the error between the model's predicted value and the true value. RMSE is calculated by taking the square root of the average sum of the squared prediction errors.

[0126] Specifically, let the model's predicted value for the i-th sample in the sample data set be h(x i ), the true value is y i , the number of samples is m, then the root mean square error can be calculated as follows:

[0127]

[0128] The smaller the RMSE, the more accurate the model prediction results.

[0129] Mean Absolute Scaled Error (MASE) is an indicator used to measure the accuracy of time series forecasting models. It is an improvement on the Mean Absolute Error (MAE) and can solve the problem that MAE cannot compare different time series well. The formula for calculating the absolute percentage error is as follows:

[0130]

[0131] Where A represents the actual observation sequence, F represents the forecast sequence, and MAE represents the mean absolute error between the historical observations and the forecast. A smaller MASE value indicates a higher predictive accuracy. When MASE equals 1, the model's forecast is the same as the historical average, indicating no predictive advantage.

[0132] The energy consumption model evaluation indicators of the dehumidification unit, circulating water pump and circulating fan are all close to 0. The comparison results are as follows: Figure 2-5 shown.

[0133] The above content is only for explaining the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A Markov-based parameter identification method for underground building internal energy consumption model, characterized in that: The following steps are involved: S1. Based on the operating conditions of the air conditioning and dehumidification units, circulating water pumps, and cooling tower fans inside the underground building and the operating characteristics of each device, energy consumption models for the air conditioning and dehumidification units, circulating water pumps, and cooling tower fans are established respectively; The energy consumption of air conditioning and dehumidification units is mainly affected by the cooling water inlet temperature T wi and outlet water temperature T wo , dehumidifier unit humidity RH and air supply temperature T ai and return air temperature T ao Therefore, the energy consumption model of the dehumidifier unit is expressed as follows P=f(RH,T wo ,T wi ,T ao ,T ai ) (1) For the convenience of operation, the difference between the inlet and outlet temperatures of the cooling water ΔTw is used. 与 Supply and return air temperature difference ΔT a calculate, ΔT w =T wo -T wi (2) ΔT a =T ai -T ao (3) Formulas (1), (2), and (3) are fitted into the following quadratic polynomial: P=x1+x2*ΔT w +x3*RH+x4*△T a +x5*ΔT w 2 +x6*RH 2 +x7*ΔT a 2 +x8*△T w *RH+x9*△T w *ΔT a +x 10 *RH*ΔT a (4) In the above formula, x1~x 10 are the parameters to be identified for the air conditioning and dehumidification unit; Similarly, according to the characteristic relationship between the circulating water pump head and flow at non-rated speed, In the above formula, c0~c2, d0~d2 are the parameters to be identified of the circulating water pump; m cw is the current circulating water pump flow, H cw is the current circulating water pump head, η cw is the working point efficiency of the circulating water pump, n is the working point speed, n nom is the rated speed; Therefore, the mathematical expression of the circulating water pump energy consumption model is: Where g d is the flow head coefficient; In step S1, the wind pressure and efficiency of the cooling tower fan can be approximately expressed as a quadratic function related to the air volume: Where, F is the wind pressure of the cooling tower fan; η is the efficiency of the cooling tower fan; Q f is the cooling tower fan air volume, m 3 / h; l1, l2, l3, l4, l5, and l6 are the parameters to be identified for the cooling tower fan. Therefore, the power consumption P of the cooling tower fan can be expressed as the following function: P=Q f *F / (3600*1000*η) (12) S2. Determine the form of the Markov model corresponding to each device, including the dimension of the state vector, the dimension of the input vector, the shape of the state transition matrix, and the input matrix; S3. Collect input data required for model parameter identification of each device, where the input data includes a state sequence of each set of state vectors and a corresponding observation value sequence; S4. Parameter estimation: For a given input data and model, the least squares method is used to estimate the model parameters and calculate the parameters to be identified for each device model; S5. Model verification and evaluation: the method is as follows: input the identified model parameters into the model in step S1 for calculation, draw a comparison chart between the calculation results and the actual measurement results, and then evaluate the performance of the model.

2. The Markov-based parameter identification method for underground building internal energy consumption model according to claim 1, characterized in that: In step S2, the dimension of the state vector, the dimension of the input vector, the shape of the state transfer matrix and the input matrix are selected according to the model properties and actual operation conditions; Specifically, in the energy consumption model of the air-conditioning and dehumidification unit, the number of parameters to be identified in the model is 10, so the dimension of the state vector and the dimension of the input vector in the corresponding Markov model are both 10; the state transfer matrix is ​​in the form of a 10*10 matrix, and the input matrix is ​​in the form of a 10*1 matrix; the number of parameters to be identified in the circulating water pump model is 6, so the dimension of the state vector and the dimension of the input vector of the circulating water pump model are both 6, the state transfer matrix is ​​in the form of a 6*6 matrix, and the input matrix is ​​in the form of a 6*1 matrix; the number of parameters to be identified in the cooling tower fan model is 6, so the dimension of the state vector and the dimension of the input vector of the cooling tower fan model are both 6, the state transfer matrix is ​​in the form of a 6*6 matrix, and the input matrix is ​​in the form of a 6*1 matrix.

3. The Markov-based parameter identification method for underground building internal energy consumption model according to claim 1, characterized in that: In step S3, in the energy consumption model of the air-conditioning and dehumidification unit, the observation value sequence is determined based on the historical operating energy consumption data of the dehumidification unit; in the circulating water pump model, the observation value sequence is determined based on the historical operating energy consumption data of the circulating water pump; in the cooling tower fan model, the observation value sequence is determined based on the historical operating energy consumption data of the cooling tower fan.

4. The Markov-based parameter identification method for underground building internal energy consumption model according to claim 1, characterized in that: In step S5, the evaluation indicators include root mean square error and absolute percentage error.

Images