Cabin air conditioner prediction method based on bidirectional interaction Mama with independent channels

Abstract

The invention relates to a cabin air conditioner prediction method based on two-way interaction Mama with independent channels, and belongs to the technical field of intelligent power utilization. According to the method, a CIBG-Mamba-interactive cycle mechanism model is constructed, and the CIBG-Mamba-interactive cycle mechanism model comprises a CIBG-Mamba and an interactive recursion mechanism; wherein the CIBG-Mama captures local modes and changes of time sequence data of different dimensions by integrating a Mama model into a recursion unit with an independent channel; according to the interactive recursion mechanism, cross-channel information interaction is achieved through a state transition function of a specific channel, and the complex dependency relationship among different dimensions is captured. The method can improve the temperature prediction precision.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent electricity consumption and relates to a cabin air conditioning prediction method based on channel-independent bidirectional interactive Mamba. Background Art

[0002] The cabin air conditioning system is a key component of new energy vehicles, playing an important role in ensuring passenger comfort and optimizing energy utilization. These systems face multi-dimensional challenges such as temperature fluctuations, humidity changes, and passenger preferences. If they cannot be effectively predicted and controlled, it may lead to a decline in passenger experience, increased battery energy consumption, and even affect driving safety.

[0003] Accurately predicting the temperature setting value of the cabin air conditioner has a profound impact on the overall performance and user experience of new energy vehicles. Through accurate prediction, intelligent temperature regulation can be achieved, energy efficiency can be improved, battery life can be extended, passenger comfort can be enhanced, and decision-making basis can be provided for vehicle management systems, thereby promoting the sustainable development of the automotive industry.

[0004] Traditional prediction models have limitations in processing multidimensional time series data of cabin air conditioning. Traditional cabin air conditioning systems are weak in processing multidimensional nonlinear data and long-term time series dependencies, with low prediction accuracy and poor adaptability to temperature changes. At the same time, the control strategy is relatively fixed and lacks dynamic adaptability, making it difficult to achieve personalized adjustment and optimal energy consumption balance. Therefore, it is urgent to design a new prediction model to improve the prediction accuracy of cabin air conditioning temperature. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a cabin air conditioning prediction method based on channel-independent two-way interactive Mamba, which aims at the limitations of traditional prediction models in processing multi-dimensional time series data of cabin air conditioning, effectively solves the shortcomings of the existing technology in processing complex multi-dimensional time series data, and significantly improves the prediction accuracy and model performance.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A cabin air conditioning prediction method based on channel-independent bidirectional interactive Mamba specifically comprises the following steps:

[0008] S1: Data acquisition and preprocessing: First, real vehicle data is acquired from multiple vehicles, mainly including characteristic data such as ambient temperature, dew point, vehicle speed and vehicle interior temperature; then the acquired raw data is cleaned;

[0009] S2: Construct a CIBG-Mamba-interactive loop mechanism model, including CIBG-Mamba and interactive recursive mechanism (IRM); CIBG-Mamba stands for channel-independent bidirectional gated Mamba, which captures local patterns and changes of time series data of different dimensions by integrating the Mamba model into the channel-independent recursive unit; the interactive recursive mechanism (IRM) realizes cross-channel information interaction through the state transfer function of a specific channel and captures the complex dependencies between different dimensions;

[0010] S3: Input the cleaned data into the CIBG-Mamba-interaction loop mechanism model to predict the user's driving setting temperature at the future time T.

[0011] Furthermore, in step S1, the acquired raw data is cleaned, and the specific steps include: a. dividing a large number of data sets into blocks for subsequent processing; b. processing each data block, and filling the upper and lower time window data of the null value data according to the timestamp; c. splicing and arranging the filled data blocks to obtain a complete data set after cleaning.

[0012] Further, in step S2, the structure of the CIBG-Mamba is to integrate the cycle unit with Mamba in a channel-independent and bidirectional manner, specifically including:

[0013] First, channel-independent embedding is used to extract channel-independent features of multidimensional time series data while retaining the unique attributes of each time series variable; this process starts from an input tensor U of shape (B, L, D) in Starting from, where B is the batch size, L is the sequence length, and D is the number of channels; the channel-independent embedding applies a slice-wise transformation to independently extract the local temporal features of each channel by the following formula:

[0014] U E =Linear(Unfold(U in ))

[0015] Unfold(·) represents the patch operation, which converts the input tensor U in Transform to U p ∈(B,S,D,P), where each channel is processed through a separate S patches, each of size P; the function further applies a channel-specific learnable linear transformation to the output; then, to obtain an independent pattern for each channel, U E Convert to U T ;

[0016] Secondly, CIBG-Mamba is used to dynamically process U T, extracting hidden features by calculating update gated Mamba, reset gated Mamba and candidate activation Mamba; this helps to better retain and utilize information over a longer period of time, improving the model's ability to learn fine temporal representations and handle long-term dependencies. In particular, the present invention forwards and reverses input U T into CIBG-Mamba for comprehensive feature learning; the CIBG-Mamba unit operates on each input in both directions at time step T as follows:

[0017] Mamba is a selective state space model with a dynamic selection mechanism; this innovative feature enables Mamba to adjust parameters based on input data, greatly improving its ability to model complex time series data and provide accurate forecasts. In traditional state space models (SSMs), the data names of the system are expressed as first-order differentials:

[0018] h' t =Ah t +Bx t

[0019] y t =Ch t

[0020] Among them, x t is the input sequence at time t, h t is the hidden state, h' t is the first-order differential representation of the hidden state, y t is the output; matrices A, B, and C are learnable parameters, and the model maintains a linear time-varying structure; in order to process continuous data, the model discretizes the sequence using the time step Δ, as shown in the following formula:

[0021]

[0022] y t =Ch t

[0023] in, I represents the identity matrix; this discretization method allows efficient computation using linear recursive methods. Mamba takes this model a step further and takes A, B, and C as input functions, as shown below:

[0024] h t =A(x t )h t-1 +B(x t )x t

[0025] y t =C(x t )ht

[0026] Where A(·), B(·) and C(·) are the input x t The state transfer function, linear projection is the operation tool used in this case.

[0027] Further, in step S2, the network structure of Mamba is:

[0028] (1) Update Gate Mamba: Update Gate A specialized Mamba module is used to determine how much of the previously hidden state H should be retained in the new state in the following formula t-1 :

[0029]

[0030] Where σ(·) is the sigmoid function, Mamba z (·) is the Mamba operation of the update gate, is the input, is the update gate, and the arrows → and ← represent the forward and backward processes respectively;

[0031] (2) Reset gate control Mamba: Reset gate U R The calculation is performed through its specific Mamba module to adjust the influence of the past state on the candidate state, as shown in the following formula:

[0032]

[0033] Among them, Mamba r It is the Mamba operation to reset the door;

[0034] (3) Candidate Activation Mamba: This component merges the current input with the previous state adjusted as follows:

[0035]

[0036] in, is a candidate activation, tanh(·) represents the hyperbolic tangent function, Mamba h is a candidate activation Mamba operation, ⊙ represents element-wise multiplication;

[0037] (4) Hidden state update: The final state is a composite of the old state and the new candidate activation state in the forward and reverse directions, where the hidden state updates in the forward and reverse directions are shown in the following formulas:

[0038]

[0039] Among them, H tis the hidden layer output.

[0040] Further, in step S2, the interactive recursive mechanism includes forward interaction and backward interaction, by adding forward and backward hidden layer outputs and The recursive interactions that form the overall information flow are as follows:

[0041] Forward interaction: The forward hidden state of the previous time step With the current backward hidden state H t Combined, we get an interactive forward hidden state, as shown in the following formula:

[0042]

[0043] Among them, Interactive(·,·) means linearly projecting the two inputs to overfit the dimension of the data stream;

[0044] Reverse interaction: reverse the hidden state of the next time step With the current forward hidden state Combined, the interactive reverse hidden state is created in the following formula:

[0045]

[0046] The above interaction state is used to calculate the update gates in both directions Reset Gate U R and candidate activation Thereby enhancing the temporal processing capability of the model; after the interactive loop mechanism, we can obtain the information of the Mamba enhanced independent loop unit in the bidirectional interactive channel, whose forward and backward hidden states imply rich temporal features, so we can obtain the final output H after the interactive loop mechanism unit is processed by the following formula out ;

[0047]

[0048] Among them, drop means dropout, convert means conversion function, and norm means layer normalization;

[0049] The interactive loop mechanism combines the bidirectional information to reconstruct the temporal information and create a time series; then, this sequence is mapped and transformed into the final output U out , as shown in the following formula:

[0050] U out =Projection(H out )

[0051] Among them, Projection represents a linear mapping.

[0052] The beneficial effects of the present invention are as follows: the present invention constructs a channel-independent bidirectional gated Mamba model (CIBG-Mamba-interactive loop mechanism) of an interactive recursive mechanism, which captures global patterns and local changes through CIBG-Mamba, and integrates cross-channel and multi-scale features using an interactive recursive mechanism, thereby improving the temperature prediction accuracy. The model of the present invention can effectively solve the deficiencies of the prior art in processing complex multidimensional time series data, and significantly improve the prediction accuracy and model performance. Specifically, it is embodied in the following aspects:

[0053] (1) Channel-Independent Bidirectional Gate-Mamba (CIBG-Mamba) Structure: Aiming at the limitations of the Mamba model in the application of multi-dimensional time series data, the present invention proposes the CIBG-Mamba structure. This structure integrates the Mamba model into channel-independent recursive units, each of which is responsible for processing time series data of a specific dimension. With this design, the model can better learn and represent fine-grained time series features and enhance the ability to capture complex time patterns.

[0054] (2) Interactive recursive mechanism (interactive loop mechanism): This paper designs an interactive recursive mechanism that cleverly integrates the historical hidden state of the forward information and the immediate hidden state of the backward information (and vice versa) through the state transition function of a specific channel. This design enables the model to achieve cross-channel information interaction and effectively capture the complex dependencies between different dimensions.

[0055] (3) CIBG-Mamba-interactive loop mechanism model: Based on the CIBG-Mamba structure and interactive recursive mechanism, the present invention constructs a complete CIBG-Mamba-interactive loop mechanism model. This model achieves high-precision prediction of multidimensional time series data by integrating the above two key components.

[0056] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0058] Figure 1 It is a structural diagram of the CIBG-Mamba-interaction cycle mechanism model of the present invention;

[0059] Figure 2 Schematic diagram of classification of input data for the model;

[0060] Figure 3 This is the model architecture diagram of Mamba;

[0061] Figure 4 All predictions for a vehicle;

[0062] Figure 5 The figure is a comparison test effect diagram of the model of the present invention and various existing models. DETAILED DESCRIPTION

[0063] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0064] See also Figure 1 to Figure 5 The present invention provides a cabin air conditioning prediction method based on channel-independent bidirectional interactive Mamba, and constructs a CIBG-Mamba-interactive cycle mechanism model (such as Figure 1 As shown in Figure 2, CIBG-Mamba fully utilizes its ability to capture global patterns and local changes in multidimensional time series data, and effectively integrates cross-channel and multi-scale features through an interactive recursive mechanism, significantly improving the utilization and prediction accuracy of multidimensional time series data. The model performs well in long-sequence data modeling and fine-grained time representation, and effectively solves the key limitations of the existing Mamba model in multidimensional time series prediction.

[0065] The cabin air conditioning prediction method based on channel-independent bidirectional interactive Mamba provided by the present invention specifically includes the following steps:

[0066] S1: Data acquisition and preprocessing: First, real vehicle data is obtained from multiple vehicles, mainly including characteristic data such as ambient temperature, dew point, vehicle speed, and vehicle interior temperature. The acquired raw data is cleaned. The specific steps include: a. Divide a large number of data sets into blocks for subsequent processing; b. Process each data block and fill the upper and lower time window data for the empty value data according to the timestamp; c. Splice and organize the filled data blocks to obtain a complete cleaned data set.

[0067] S2: Data input and feature construction: Input the cleaned data into the CIBG-Mamba-interaction loop mechanism model. Figure 2 As shown in the figure, the input feature data can be divided into three categories: a. External features (blue boxes): such as ambient temperature, dew point, etc.; b. Internal features (cyan boxes): such as internal temperature, etc.; c. User-set features (red boxes): such as target temperature set by the user, etc. The constructed input data format is (B, L, D), where B represents the batch size, L represents the input sequence length, and D represents the feature dimension.

[0068] S3: Model prediction: The CIBG-Mamba-interactive recurrent mechanism model of the present invention is mainly composed of two key components: a. CIBG-Mamba: by integrating the Mamba model into the channel-independent recurrent unit, it captures the local patterns and changes of time series data of different dimensions; b. Interactive recurrent mechanism (IRM): through the state transfer function of a specific channel, it realizes cross-channel information interaction and captures the complex dependencies between different dimensions. The specific structure of the model is as follows Figure 1 The prediction procedure is shown in Table 1.

[0069] Table 1 Prediction algorithm flow

[0070]

[0071]

[0072] S4: Prediction result output: Through the CIBG-Mamba-interaction loop mechanism model, the user's driving setting temperature at the future time T can be predicted. The output format is (B, T, D), where B represents the batch size, T represents the predicted number of future time steps, and D represents the predicted feature dimension. This algorithm mainly predicts the temperature set by the user.

[0073] 1) CIBG-Mamba: Channel Independent Bidirectional Gated Mamba

[0074] CIBG-Mamba enhances multidimensional time series data modeling by integrating recurrent units with Mamba in a channel-independent and bidirectional manner. This integration improves information flow control by using a method designed specifically for multidimensional time series prediction that is more effective than traditional gating mechanisms.

[0075] First, channel-independent embedding is used to extract channel-independent features of multidimensional time series data while retaining the unique attributes of each time series variable. This process starts with an input tensor U of shape (B, L, D) in Starting from, where B is the batch size, L is the sequence length, and D is the number of channels. CIE applies a slice-wise transformation to extract local temporal features for each channel independently by the following formula:

[0076] U E =Linear(Unfold(U in ))

[0077] Unfold(·) represents the patch operation, which converts the input tensor U in Transform to U p ∈(B,S,D,P), where each channel is passed through a separate S patch, each patch size is P. The function further applies a channel-specific learnable linear transformation to the output. Then, in order to obtain the independent mode of each channel, the present invention converts to U T .

[0078] Secondly, CIBG-Mamba is proposed to dynamically process U T , extracting hidden features by calculating update gated Mamba, reset gated Mamba, and candidate activation Mamba. This helps to better retain and utilize information over a longer period of time, improving the model's ability to learn fine-grained temporal representations and handle long-term dependencies. In particular, we forward and backward input U T into CIBG-Mamba for comprehensive feature learning. The CIBG-Mamba unit operates on each input in both directions at time step T as follows:

[0079] Mamba is a selective state space model with a dynamic selection mechanism. This innovative feature enables Mamba to adjust parameters based on input data, greatly improving its ability to model complex time series data and provide accurate forecasts. In traditional state space models (SSMs), the data names of the system are expressed as first-order differentials:

[0080] h' t =Ah t +Bx t

[0081] y t =Ch t

[0082] Among them, x t is the input sequence at time t, h t is the hidden state, h' t is the first-order differential representation of the hidden state, y t is the output; matrices A, B, and C are learnable parameters, and the model maintains a linear time-varying structure; in order to process continuous data, the model discretizes the sequence using the time step Δ, as shown in the following formula:

[0083]

[0084] y t =Ch t

[0085] in, This discretization method allows efficient computation using linear recursive methods. Mamba takes this model a step further and takes A, B, and C as input functions, as shown below:

[0086] h t =A(x t )h t-1 +B(x t )x t

[0087] y t =C(x t )h t

[0088] Where A(·), B(·) and C(·) are the input x t The state transfer function, linear projection is the operation tool used in this case. The network structure of Mamba is as follows Figure 3 shown.

[0089] (1) Update Gate Mamba: Update Gate A specialized Mamba module is used to determine how much of the previously hidden state H should be retained in the new state in the following formula t-1 :

[0090]

[0091]

[0092] Where σ(·) is the sigmoid function, and the arrows → and ← represent the forward and backward processes, respectively;

[0093] (2) Reset gate control Mamba: Reset gate U R The calculation is performed through its specific Mamba module to adjust the influence of the past state on the candidate state, as shown in the following formula:

[0094]

[0095] (3) Candidate Activation Mamba: This component merges the current input with the previous state adjusted as follows:

[0096]

[0097] Where tanh(·) represents the hyperbolic tangent function, and ⊙ represents element-wise multiplication;

[0098] (4) Hidden state update: The final state is a composite of the old state and the new candidate activation state in the forward and reverse directions, where the hidden state updates in the forward and reverse directions are shown in the following formulas:

[0099]

[0100] 2) Interactive loop mechanism

[0101] The interactive recursive mechanism includes forward interaction and backward interaction, by adding forward and backward hidden layer outputs and The recursive interactions that form the overall information flow are as follows:

[0102] Forward interaction: The forward hidden state of the previous time step With the current backward hidden state Combined, we get an interactive forward hidden state, as shown in the following formula:

[0103]

[0104] Among them, Interactive(·,·) means linearly projecting the two inputs to overfit the dimension of the data stream;

[0105] Reverse interaction: reverse the hidden state of the next time step With the current forward hidden state Combined, the interactive reverse hidden state is created in the following formula:

[0106]

[0107] The above interaction state is used to calculate the update gates in both directions Reset Gate U R and candidate activation Thereby enhancing the time processing capability of the model; after the interactive loop mechanism, the present invention can obtain the information of the Mamba enhanced independent loop unit in the two-way interactive channel, and its forward and backward hidden states imply rich temporal features. Therefore, the present invention can obtain the final output H after the interactive loop mechanism unit is processed by the following formula out ;

[0108]

[0109] Among them, drop means dropout, convert means conversion function, and norm means layer normalization.

[0110] The interactive loop mechanism combines the bidirectional information to reconstruct the temporal information and create a time series; then, this sequence is mapped and transformed into the final output U out, as shown in the following formula:

[0111] U out =Projection(H out )

[0112] Experimental results analysis:

[0113] For the final results of temperature setting, this experiment was conducted 30 times, and the final MSE and MAE mean and standard deviation results were compared with the existing most advanced algorithms, as shown in Table 2.

[0114] Table 2 Performance comparison of different models

[0115]

[0116] In Table 2, the italics indicate the best results, and the bold black indicates the second best results. It can be seen that the model we proposed has excellent performance in both prediction accuracy and prediction precision. The prediction accuracy is the highest among all models, with the mean MSE of 0.904 and the mean MAE of 0.641. The prediction accuracy ranks second among all models, with the prediction standard deviation of MSE of 0.057 and the prediction standard deviation of MAE of 0.038.

[0117] This experiment visualizes all the predictions for a car, such as Figure 4 As shown in the figure, compared with other models, the predicted value of the model of the present invention is closer to the true value, indicating that its performance is better than other models. In addition, the present invention also enlarges the fluctuation points in the figure, which are caused by the rapid changes of users and the environment. It can be seen from the enlarged sub-graph that the prediction curve of the model of the present invention has a larger slope and can produce faster and more accurate adaptation results. This phenomenon further proves that the model proposed by the present invention has a fine time representation capability.

[0118] In order to prove the more refined time representation capability of the model proposed in this invention, we conducted comparative experiments on more models, such as Figure 5 As shown, it can be seen that when the time fluctuation lasts for dozens of data points, the algorithm proposed in the present invention obtains the most accurate prediction result and the fastest convergence ability.

[0119] It can be seen that the CIBG-Mamba-interaction loop mechanism model has the following significant advantages:

[0120] (1) High prediction accuracy: By cleverly combining the CIBG-Mamba structure with the interactive recursive mechanism, this model can make full use of the rich information contained in multidimensional time series data, perform modeling and prediction from both global and local perspectives, and significantly improve the prediction accuracy of the model.

[0121] (2) Strong feature capture capability: The CIBG-Mamba structure enables the model to capture fine-grained time series features. The independent recursive unit of each channel can focus on processing time series data of a specific dimension. At the same time, the interactive recursive mechanism realizes cross-dimensional information interaction, enabling the model to capture complex dependencies between different dimensions.

[0122] (3) Ability to process complex data: The CIBG-Mamba-interactive recurrent mechanism model is designed to process long-sequence, high-dimensional complex time series data. By introducing channel-independent recursive units and interactive recursive mechanisms, this model can effectively deal with the high-dimensional characteristics and long-term dependency problems of data, and has broad application prospects in fields involving complex time series such as climate forecasting and financial analysis.

[0123] In summary, the CIBG-Mamba-interactive cycle mechanism model of the present invention has significant innovations and advantages in prediction accuracy, feature capture capability and processing of complex data, and provides a new solution and method for multidimensional time series prediction tasks, which has important theoretical significance and practical value.

[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A cabin air conditioning prediction method based on channel-independent bidirectional interactive Mamba, characterized in that: The method specifically comprises the following steps: S1: Data acquisition and preprocessing: First, real vehicle data is acquired from multiple vehicles, including ambient temperature, dew point, vehicle speed, and vehicle interior temperature; then the acquired raw data is cleaned; S2: Construct a CIBG-Mamba-interactive loop mechanism model, including CIBG-Mamba and interactive recursive mechanism; CIBG-Mamba stands for channel-independent bidirectional gated Mamba, which captures local patterns and changes of time series data of different dimensions by integrating the Mamba model into the channel-independent recursive unit; the interactive recursive mechanism realizes cross-channel information interaction through the state transfer function of a specific channel, and captures the complex dependencies between different dimensions; S3: Input the cleaned data into the CIBG-Mamba-interaction loop mechanism model to predict the user's driving setting temperature at the future time T.

2. The cabin air conditioning prediction method based on channel-independent bidirectional interactive Mamba according to claim 1 is characterized in that: In step S1, the acquired raw data is cleaned, and the specific steps include: a. dividing a large number of data sets into blocks; b. processing each data block, and filling the upper and lower time window data of the null value data according to the timestamp; c. splicing and arranging the filled data blocks to obtain a complete data set after cleaning.

3. The cabin air conditioning prediction method based on channel-independent bidirectional interactive Mamba according to claim 2 is characterized in that: In step S2, the structure of CIBG-Mamba specifically integrates the cycle unit with Mamba in a channel-independent and bidirectional manner, specifically including: First, channel-independent embedding is used to extract channel-independent features of multidimensional time series data while retaining the unique attributes of each time series variable; this process starts from an input tensor U of shape (B, L, D) in Starting from, where B is the batch size, L is the sequence length, and D is the number of channels; the slice transformation is applied to extract the local temporal features of each channel independently by the following formula: U E =Linear(Unfold(U in )) Unfold(·) represents the patch operation, which converts the input tensor U in Transform to U p ∈(B,S,D,P), where each channel is passed through a separate S patch, each patch size is P; in order to obtain an independent pattern for each channel, U E Convert to U T ; Secondly, CIBG-Mamba is used to dynamically process U T , extract hidden features by calculating update gated Mamba, reset gated Mamba and candidate activation Mamba; forward and reverse input U T into CIBG-Mamba for comprehensive feature learning; the CIBG-Mamba unit operates on each input in both directions at time step T as follows: In the traditional state-space model, the data of the system is expressed as a first-order differential: h' t =Ah t +Bx t the t =Ch t Among them, x t is the input sequence at time t, h t is the hidden state, h' t is the first-order differential representation of the hidden state, y t is the output; matrices A, B, and C are learnable parameters, and the model maintains a linear time-varying structure; in order to process continuous data, the model discretizes the sequence using the time step Δ, as shown in the following formula: the t =Ch t in, I represents the identity matrix; Mamba takes this model a step further and uses A, B, and C as input functions, as shown below: h t =A(x t )h t-1 +B(x t )x t y t =C(x t )h t Where A(·), B(·) and C(·) are the input x t The state transfer function, linear projection is the operation tool used in this case.

4. The cabin air conditioning prediction method based on channel-independent bidirectional interactive Mamba according to claim 3 is characterized in that: In step S2, the network structure of Mamba is: (1) Update Gate Mamba: Update Gate A specialized Mamba module is used to determine how much of the previously hidden state H should be retained in the new state in the following formula t-1 : Where σ(·) is the sigmoid function, Mamba z (·) is the Mamba operation of the update gate, is the input, is the update gate, and the arrows → and ← represent the forward and backward processes respectively; (2) Reset gate control Mamba: Reset gate U R The calculation is performed through its specific Mamba module to adjust the influence of the past state on the candidate state, as shown in the following formula: Among them, Mamba r It is the Mamba operation to reset the door; (3) Candidate Activation Mamba: Merge the current input with the adjusted previous state as follows: in, is a candidate activation, tanh(·) represents the hyperbolic tangent function, Mamba h is the Mamba operation of the candidate activation gate, ⊙ Represents element-wise multiplication; (4) Hidden state update: The final state is a composite of the old state and the new candidate activation state in the forward and reverse directions, where the hidden state updates in the forward and reverse directions are shown in the following formulas: Among them, H t is the hidden layer output.

5. The cabin air conditioning prediction method based on channel-independent two-way interactive Mamba according to claim 4 is characterized in that: In step S2, the interactive recursive mechanism includes forward interaction and backward interaction, by adding forward and backward hidden layer outputs and The recursive interactions that form the overall information flow are as follows: Forward interaction: The forward hidden state of the previous time step With the current backward hidden state Combined, we get an interactive forward hidden state, as shown in the following formula: Among them, Interactive(·,·) means linearly projecting the two inputs to fit the dimension of the data stream; Reverse interaction: reverse the hidden state of the next time step With the current forward hidden state Combined, the interactive reverse hidden state is created in the following formula: The above interaction state is used to calculate the update gates in both directions Reset Gate U R and candidate activation Thereby enhancing the temporal processing capability of the model; after the interactive loop mechanism, the information of the Mamba enhanced independent loop unit in the bidirectional interactive channel is obtained, and its forward and backward hidden states imply rich temporal features, so the final output H of the interactive loop mechanism unit after processing by the following formula is obtained out ; Among them, drop means dropout, convert means conversion function, and norm means layer normalization; The interactive loop mechanism combines the bidirectional information to reconstruct the temporal information and create a time series; then, this sequence is mapped and transformed into the final output U out , as shown in the following formula: U out =Projection(H out ) Among them, Projection represents a linear mapping.

Images