Energy storage container temperature control method based on LSTM
Through the LSTM-based energy storage container temperature control method, the heating or cooling power is automatically adjusted using long and short-term memory networks and temperature models, the problem that traditional temperature control systems are difficult to cope with temperature changes in complex environments in high altitude and high cold areas is solved, and precise control and energy conservation are achieved.
Patent Information
- Application Number
- CN202510157626.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-13
- Publication Date
- 2025-05-16
AI Technical Summary
Traditional temperature control systems are difficult to effectively deal with temperature changes in complex environments in high altitude and high cold areas, and there are problems such as low energy efficiency, slow response, large temperature fluctuations and insufficient adaptability.
The LSTM-based energy storage container temperature control method is adopted to predict outdoor temperature for a long time by building a long-term memory network, and combine the temperature model and control rate model in the container to automatically adjust the heating or cooling power to ensure that the indoor temperature is within the optimal range.
It realizes precise control of indoor temperature, reduces energy waste, improves the adaptability and response speed of the temperature control system, and adapts to the harsh environment of high-altitude areas.
Smart Images
Figure CN120010593A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of energy storage technology, and in particular to a temperature control method for an energy storage container based on LSTM. Background Art
[0002] The climate characteristics of high-altitude and cold regions are large temperature differences between day and night, extremely low temperatures in winter, and extreme weather that easily affects the indoor environment. Therefore, traditional temperature control systems in such areas often face problems such as low energy efficiency, slow response, and large temperature fluctuations. In order to improve the stability and accuracy of the temperature control system, it is necessary to combine modern artificial intelligence technology for intelligent prediction and regulation.
[0003] With the continuous development of infrastructure construction in high-altitude areas, the temperature control requirements for storage equipment are getting higher and higher. Temperature control is particularly important in high-altitude areas.
[0004] At present, the existing temperature control technology still has many defects when dealing with temperature changes in complex and cold environments. Traditional temperature control systems lack sufficient adaptability and cannot effectively respond to sudden changes in the external environment, resulting in low temperature control accuracy and slow response speed. At the same time, these systems often do not take into account multi-dimensional environmental factors, which makes temperature regulation unstable and inefficient. In addition, traditional control methods consume a lot of energy and lack optimization of energy efficiency, resulting in more serious energy waste problems in cold areas. Furthermore, existing technologies are also insufficient in fault diagnosis, long-term prediction and optimization capabilities, and it is difficult to meet the needs of cold areas for high reliability and continuous operation of temperature control systems. Summary of the invention
[0005] The purpose of the present invention is to provide a temperature control method for an energy storage container based on LSTM, which can automatically adjust the heating or cooling power according to the external environment and the temperature difference between the inside and outside through the optimization control of the control rate model, ensure that the indoor temperature is within the optimal range, and effectively reduce energy waste.
[0006] A temperature control method for an energy storage container based on LSTM, comprising:
[0007] Constructing a long short-term memory network LSTM, training the LSTM, and obtaining the trained LSTM;
[0008] Obtain outdoor temperature data before the current moment, and input the outdoor temperature data into the trained LSTM to obtain outdoor time series temperature prediction data after the current moment;
[0009] Constructing a temperature model inside the container, and obtaining time series temperature prediction data inside the container based on the temperature model inside the container and the outdoor time series temperature prediction data; the temperature model inside the container includes a heat conduction model, a convection heat transfer model and a radiation heat transfer model;
[0010] Constructing a control rate model, inputting the time series temperature prediction data in the container into the control rate model, and obtaining the output of the controller with the minimum value of the control rate model as the goal;
[0011] The control rate model expression is:
[0012]
[0013] Where: T in (t) is the predicted temperature value in the container at time t, T setpoint is the target temperature, μ(t) is the output of the controller at time t, λ is the penalty coefficient, N is the prediction window length, and J is the control rate model.
[0014] Optionally, the temperature model expression in the container is:
[0015]
[0016] Where: Q cond is the heat conduction model, Q conv is the convection heat transfer model, Q rad is the radiation heat transfer model, C is the heat capacity of the object inside the container, T in is the predicted temperature inside the container at time k.
[0017] Optionally, the heat conduction model expression is:
[0018]
[0019] Where: k is the thermal conductivity of the container, A is the area of the container wall, ΔT is the temperature difference between the inside and outside of the container, ΔT is the difference between the predicted outdoor temperature at time k and the predicted indoor temperature at time k-1, the predicted outdoor temperature is based on the outdoor time series temperature prediction data, and Δx is the thickness of the heat conduction path.
[0020] Optionally, the convective heat transfer model expression is:
[0021] Q conv =hA(T out -T in );
[0022] Where: h is the convection heat transfer coefficient of the container, T out is the predicted value of outdoor temperature at time k, and A is the area of the container wall.
[0023] Optionally, the radiation heat transfer model expression is:
[0024]
[0025] Where: ε is the emissivity of the container surface, σ is the Stefan-Boltzmann constant.
[0026] Optionally, the control rate model is constructed, the time series temperature prediction data in the container is input into the control rate model, and the output of the controller is obtained with the minimum value of the control rate model as the goal, specifically:
[0027] Constructing the control rate model and the constraint model, taking the constraint model as a restriction condition, inputting the time series temperature prediction data in the container into the control rate model, and obtaining the output of the controller with the minimum value of the control rate model as the goal;
[0028] The constraint model expression is:
[0029]
[0030] Where: T min is the minimum temperature in the container, T max is the maximum temperature inside the container.
[0031] Optionally, the LSTM includes a forget gate, an input gate, a candidate memory unit, an update memory unit, an output gate and a hidden state.
[0032] Optionally, the forget gate expression is: t =σ(W f .[h t-1 ,X t ]+b f );
[0033] The input gate: i t =σ(W i .[h t-1 ,X t ]+b i );
[0034] The candidate memory unit:
[0035] The updating memory unit:
[0036] The output gate: o t =σ(W o .[h t-1 ,X t ]+bo );
[0037] The hidden state: h t =o t *tanh(C t );
[0038] Where: f t is the output of the forget gate of the t-th network unit, σ and tanh are activation functions, b f is the bias of the forget gate, W f is the weight of the forget gate, b i is the bias of the input gate, b C is the bias of the candidate memory cell, b o is the bias of the output gate, W i is the weight of the input gate, W c is the weight of the candidate memory unit, W o is the weight of the output gate, h t is the output of the hidden state of the tth network unit, h t-1 is the output of the hidden state of the t-1th network unit, o t is the output of the output gate of the tth network unit, C t is the output of the updated memory unit of the tth network unit, X t is the input of the tth network unit, C t-1 is the output of the updated memory unit of the t-1th network unit, i t is the output of the input gate of the tth network unit, is the output of the candidate memory unit of the tth network unit.
[0039] The effects of the present invention are as follows:
[0040] The LSTM-based energy storage container temperature control method of the present invention uses LSTM to predict the long-term outdoor temperature and identify possible temperature fluctuations in advance, so that the temperature control system can be fully prepared, avoiding the problems of lag and lack of predictability of traditional systems.
[0041] The LSTM-based energy storage container temperature control method of the present invention uses digital twin technology to simulate physical processes such as heat conduction and convection inside the container, and can accurately predict the changing trend of indoor temperature and perform real-time adjustment.
[0042] The LSTM-based energy storage container temperature control method of the present invention can automatically adjust the heating or cooling power according to the external environment and the internal and external temperature difference through the optimization control of the control rate model, thereby ensuring that the indoor temperature is within the optimal range and effectively reducing energy waste.
[0043] The LSTM-based energy storage container temperature control method of the present invention can adapt to the harsh environment of high-cold and high-altitude areas, and can also maintain the stability of temperature control under extreme conditions to avoid adverse effects of indoor temperature fluctuations on batteries. It can automatically optimize the temperature control strategy according to real-time data and environmental changes, has strong adaptive capabilities, and can cope with various complex environmental conditions and operational requirements. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] Figure 1 This is a flow chart of the temperature control method of the energy storage container based on LSTM of the present invention. DETAILED DESCRIPTION
[0045] Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0046] Figure 1 This is a flow chart of the temperature control method of the energy storage container based on LSTM of the present invention. Figure 1 As shown, the present invention provides a temperature control method for an energy storage container based on LSTM, which includes:
[0047] S1, construct a long short-term memory network LSTM, train the LSTM, and obtain a trained LSTM.
[0048] Specifically, LSTM includes a forget gate, an input gate, a candidate memory cell, an update memory cell, an output gate, and a hidden state.
[0049] The expression of the forget gate is: t =σ(W f .[h t-1 ,X t ]+b f ).
[0050] Input gate: i t =σ(W i .[h t-1 ,X t ]+b i ).
[0051] Candidate memory cells:
[0052] Update memory unit:
[0053] Output gate: o t =σ(W o .[h t-1 ,X t ]+b o ).
[0054] Hidden state: h t =ot *tanh(C t ).
[0055] Where: f t is the output of the forget gate of the t-th network unit, σ and tanh are activation functions, b f is the bias of the forget gate, W f is the weight of the forget gate, b i is the bias of the input gate, b C is the bias of the candidate memory cell, b o is the bias of the output gate, W i is the weight of the input gate, W c is the weight of the candidate memory unit, W o is the weight of the output gate, h t is the output of the hidden state of the tth network unit, h t-1 is the output of the hidden state of the t-1th network unit, o t is the output of the output gate of the tth network unit, C t is the output of the updated memory unit of the tth network unit, X t is the input of the tth network unit, C t-1 is the output of the updated memory unit of the t-1th network unit, i t is the output of the input gate of the tth network unit, is the output of the candidate memory unit of the tth network unit.
[0056] During the training process, LSTM updates W f ,W i ,W C ,W o and b f ,b i ,b C ,b o , and optimize the prediction target through the loss function.
[0057] S2, obtain the outdoor temperature data before the current moment, and input the outdoor temperature data into the trained LSTM to obtain the outdoor time series temperature prediction data after the current moment.
[0058] S3, constructing a temperature model inside the container, and obtaining time series temperature prediction data inside the container based on the temperature model inside the container and the outdoor time series temperature prediction data. The temperature model inside the container includes a heat conduction model, a convection heat transfer model, and a radiation heat transfer model.
[0059] Preferably, the temperature model expression in the container is:
[0060]
[0061] Where: Q cond is the heat conduction model, Q conv is the convection heat transfer model, Q rad is the radiation heat transfer model, C is the heat capacity of the object inside the container, T in is the predicted temperature inside the container at time k.
[0062] The heat conduction model expression is:
[0063]
[0064] Where: k is the thermal conductivity of the container, A is the area of the container wall, ΔT is the temperature difference between the inside and outside of the container, ΔT is the difference between the predicted outdoor temperature at time k and the predicted indoor temperature at time k-1, the predicted outdoor temperature is based on the outdoor time series temperature prediction data, and Δx is the thickness of the heat conduction path.
[0065] The convective heat transfer model expression is:
[0066] Q conv =hA(T out -T in );
[0067] Where: h is the convection heat transfer coefficient of the container, T out is the predicted value of outdoor temperature at time k, and A is the area of the container wall.
[0068] The radiation heat transfer model expression is:
[0069]
[0070] Where: ε is the emissivity of the container surface, σ is the Stefan-Boltzmann constant.
[0071] S4, construct a control rate model, input the time series temperature prediction data in the container into the control rate model, and obtain the output of the controller with the minimum value of the control rate model as the goal.
[0072] Specifically, the temperature inside the container is adjusted by optimizing the control strategy to ensure that the temperature remains within a target range, for example, between 20°C and 30°C.
[0073] The control rate model expression is:
[0074]
[0075] Where: T in (t) is the predicted temperature value in the container at time t, T setpoint is the target temperature, μ(t) is the output of the controller at time t, λ is the penalty coefficient, N is the prediction window length, and J is the control rate model.
[0076] Specifically, a control rate model and a constraint model are constructed, and the time series temperature prediction data in the container is input into the control rate model with the constraint model as the restriction condition, and the output of the controller is obtained with the minimum value of the control rate model as the goal.
[0077] The constraint model expression is:
[0078]
[0079] Where: T min is the minimum temperature in the container, T max is the maximum temperature inside the container.
[0080] The embodiments described above are only descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.
Claims
1. A temperature control method for energy storage container based on LSTM, characterized in that: It includes: Constructing a long short-term memory network LSTM, training the LSTM, and obtaining the trained LSTM; Obtain outdoor temperature data before the current moment, and input the outdoor temperature data into the trained LSTM to obtain outdoor time series temperature prediction data after the current moment; Constructing a temperature model inside the container, and obtaining time series temperature prediction data inside the container based on the temperature model inside the container and the outdoor time series temperature prediction data; the temperature model inside the container includes a heat conduction model, a convection heat transfer model and a radiation heat transfer model; Constructing a control rate model, inputting the time series temperature prediction data in the container into the control rate model, and obtaining the output of the controller with the minimum value of the control rate model as the goal; The control rate model expression is: Where: T in (t) is the predicted temperature value in the container at time t, T setpoint is the target temperature, μ(t) is the output of the controller at time t, λ is the penalty coefficient, N is the prediction window length, and J is the control rate model.
2. The LSTM-based energy storage container temperature control method according to claim 1 is characterized in that: The temperature model expression inside the container is: Where: Q cond is the heat conduction model, Q conv is the convection heat transfer model, Q rad is the radiation heat transfer model, C is the heat capacity of the object inside the container, T in is the predicted temperature inside the container at time k.
3. The LSTM-based energy storage container temperature control method according to claim 1 is characterized in that: The heat conduction model expression is: Where: k is the thermal conductivity of the container, A is the area of the container wall, ΔT is the temperature difference between the inside and outside of the container, ΔT is the difference between the predicted outdoor temperature at time k and the predicted indoor temperature at time k-1, the predicted outdoor temperature is based on the outdoor time series temperature prediction data, and Δx is the thickness of the heat conduction path.
4. The LSTM-based energy storage container temperature control method according to claim 2 is characterized in that: The convective heat transfer model expression is: Q conv =hA(T out -T in ); Where: h is the convection heat transfer coefficient of the container, T out is the predicted value of outdoor temperature at time k, and A is the area of the container wall.
5. The LSTM-based energy storage container temperature control method according to claim 4 is characterized in that: The radiation heat transfer model expression is: Where: ε is the emissivity of the container surface, σ is the Stefan-Boltzmann constant.
6. The LSTM-based energy storage container temperature control method according to claim 4 is characterized in that: The control rate model is constructed, the time series temperature prediction data in the container is input into the control rate model, and the output of the controller is obtained with the minimum value of the control rate model as the goal, specifically: Constructing the control rate model and the constraint model, taking the constraint model as a restriction condition, inputting the time series temperature prediction data in the container into the control rate model, and obtaining the output of the controller with the minimum value of the control rate model as the goal; The constraint model expression is: Where: T min is the minimum temperature in the container, T max is the maximum temperature inside the container.
7. The LSTM-based energy storage container temperature control method according to claim 1 is characterized in that: The LSTM includes a forget gate, an input gate, a candidate memory unit, an update memory unit, an output gate and a hidden state.
8. The LSTM-based energy storage container temperature control method according to claim 7 is characterized in that: The forget gate expression is: t =σ(W f .[h t-1 ,X t ]+b f ); The input gate: i t =σ(W i .[h t-1 ,X t ]+b i ); The candidate memory unit: The updating memory unit: The output gate: o t =σ(W o .[h t-1 ,X t ]+b o ); The hidden state: h t =o t *tanh(C t ); Where: f t is the output of the forget gate of the t-th network unit, σ and tanh are activation functions, b f is the bias of the forget gate, W f is the weight of the forget gate, b i is the bias of the input gate, b C is the bias of the candidate memory cell, b o is the bias of the output gate, W i is the weight of the input gate, W c is the weight of the candidate memory unit, W o is the weight of the output gate, h t is the output of the hidden state of the tth network unit, h t-1 is the output of the hidden state of the t-1th network unit, o t is the output of the output gate of the tth network unit, C t is the output of the updated memory unit of the tth network unit, X t is the input of the tth network unit, C t-1 is the output of the updated memory unit of the t-1th network unit, i t is the output of the input gate of the tth network unit, is the output of the candidate memory unit of the tth network unit.