A photovoltaic multi-source sensing fusion and number same transmission integrated device and system

Abstract

The application relates to the field of photovoltaic technology and discloses a photovoltaic multi-source sensing fusion odd-even same-transmission integrated device and system, wherein the photovoltaic multi-source sensing fusion odd-even same-transmission integrated system comprises a sensor data synchronous transmission module, which is used for sending sensor data to a data fusion module after time synchronization; the data fusion module obtains a fusion vector by performing data fusion on the sensor data through a data fusion model; and a deviation detection module calculates the deviation value of the actual angle of a horizontal shaft and a vertical shaft of a photovoltaic panel and a control target angle through the fusion vector. Through photovoltaic multi-source sensing fusion, the application can acquire and analyze various sensor data of the photovoltaic panel and the environment, and provide a photovoltaic panel angle with high power generation efficiency and long service life. Compared with the deviation value calculated according to single sensing data, the system can balance various adjustment targets and prolong the service life of the photovoltaic panel.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic technology, and more specifically, to a photovoltaic multi-source sensor fusion heterogeneous simultaneous transmission integrated device and system. Background Technology

[0002] When using photovoltaic panels to generate electricity, the angle of the photovoltaic panels needs to be adjusted to achieve the best power generation efficiency. In order to ensure that the photovoltaic panels can absorb solar energy to the maximum extent, the angle of the sun is detected and the photovoltaic panels are adjusted to match the angle of direct sunlight. However, when the sunlight is too strong, the photovoltaic panels will be exposed to direct sunlight, which will cause the temperature to be too high, affecting the power generation efficiency, accelerating the aging of the photovoltaic panels, and shortening their service life. At this time, it is necessary to adjust the angle of the photovoltaic panels to reduce direct radiant heat, thereby controlling the temperature of the photovoltaic panels.

[0003] In extreme weather conditions, such as strong winds or heavy rain, the photovoltaic panels need to be adjusted to a safe angle to avoid physical damage. Summary of the Invention

[0004] The purpose of this invention is to provide a photovoltaic multi-source sensing fusion integrated device, system, and storage medium to solve the above-mentioned problems.

[0005] This invention provides a photovoltaic multi-source sensor fusion heterogeneous simultaneous transmission integrated system, comprising:

[0006] The sensor data synchronization and transmission module is used to synchronize sensor data in time and then send it to the data fusion module.

[0007] Sensor data is collected once at each sampling point at each time. It consists of the measured values ​​detected by the sensor, including photovoltaic data and environmental data. The photovoltaic data includes the average solar irradiance of the photovoltaic panel within a preset time period, the average photovoltaic panel temperature within a preset time period, and the current horizontal and vertical axis angles of the photovoltaic panel. The environmental data includes the current solar azimuth and elevation angles, the average solar irradiance of the environment within a preset time period, the average wind speed and wind direction within a preset time period, and the average rainfall within a preset time period.

[0008] A time sampling point refers to the moment when data is collected at predetermined time intervals, each interval being 5 minutes.

[0009] The data fusion module uses a data fusion model to fuse sensor data and obtain a fusion vector.

[0010] The deviation detection module calculates the deviation between the actual angles of the horizontal and vertical axes of the photovoltaic panel and the control target angle by using fused vectors.

[0011] Furthermore, For the first Sensor data at each time sampling point:

[0012] ;

[0013] in, The average solar irradiance of the photovoltaic panels within a preset time period. The average temperature of the photovoltaic panel within a preset time period. The current horizontal axis angle of the photovoltaic panel. The current vertical axis angle of the photovoltaic panel. This is the current solar azimuth angle. This is the current solar altitude angle. The average solar radiation intensity in the environment during a preset time period. The average wind speed over a preset time period. The average wind direction over a preset time period. This represents the average rainfall over a preset time period.

[0014] Furthermore, the data fusion model uses a Long Short-Term Memory (LSTM) network, which includes a forget gate, an input gate, cell states, and an output gate. The output of the forget gate... for:

[0015] ;

[0016] in, It is the weight matrix of the forget gate. It is the bias term of the forgetting gate. It is the hidden state of the previous time step. Sensor data at the current time step splicing, It is the Sigmoid activation function;

[0017] Input gate output With candidate cell state The calculation method is as follows:

[0018] ;

[0019] ;

[0020] in, It is the weight matrix of the input gate. It is the bias term of the input gate. It is the hyperbolic tangent activation function. It is the weight matrix for calculating the candidate cell states. It is a bias term for the candidate cell state;

[0021] Cell state It is from the previous cell state The weighted synthesis based on candidate cell states is calculated as follows:

[0022] ;

[0023] in, This represents an element-wise multiplication operation;

[0024] Activation value of output gate Hidden state at the current time step The calculation method is as follows:

[0025] ;

[0026] ;

[0027] in, It is the weight matrix of the output gate. It is the bias term of the output gate;

[0028] Data fusion model outputs fusion vector , It can be represented as:

[0029] .

[0030] Furthermore, the actual angle is In and The target angle is the action output by the decision neural network;

[0031] An action is represented as:

[0032] ;

[0033] in, They represent the first time. At each time sampling point, the angles that the horizontal and vertical axes of the photovoltaic panel need to be adjusted to.

[0034] Furthermore, a decision neural network is trained using Q-learning, and a reward function is set. This reward function needs to comprehensively consider power generation efficiency, adjustment costs, environmental impact, and the temperature effect of the photovoltaic panels. Power generation efficiency is the ratio of actual power generation to the maximum theoretical power generation, and adjustment costs are the energy consumed to adjust the photovoltaic panels. The reward function is expressed as follows:

[0035] ;

[0036] in, This represents the immediate reward at the t-th time sampling point. This represents the actual power generation of the photovoltaic panel within five minutes prior to sampling point t. This represents the maximum theoretical power generation of the photovoltaic panel within the five minutes preceding the t time sampling points. and Let represent the deviation values ​​of the adjustment range of the photovoltaic panel's horizontal and vertical axes at the t-th time sampling point, respectively. This represents the environmental impact at the t-th time sampling point, assuming the photovoltaic panels were protected under harsh environmental conditions. A value of 1 indicates that the photovoltaic panel is not protected. The value is -1, otherwise The value is 0. This represents the effect of photovoltaic panel temperature at the sampling point at time t. The value is the average photovoltaic panel temperature within the preset time period at the t-th time sampling point. The larger of the difference from the safe temperature threshold and 0. , , , These are weighting parameters, with default values ​​of 0.5, 0.2, 0.2, and 0.1.

[0037] Furthermore, the steps of Q-learning include:

[0038] Step 101, replicate the decision neural network to obtain network;

[0039] Step 102: Every fixed number of network updates, update the target network so that its parameters are the same as the parameters of the current Q network;

[0040] Step 103 continues until the decision neural network converges or the number of iterations reaches the set number, at which point the process terminates.

[0041] Furthermore, an ε-greedy strategy is adopted when selecting actions, and the methods for selecting actions include:

[0042] Step 201, calculate the first... Exploration parameters used to select actions during step value , The calculation method is as follows:

[0043] ;

[0044] in: At the start of training The initial value is 0.9, with a default value of 0.9. for The minimum achievable value, with a default value of 0.1; U is the total number of steps in the entire training process, with a default value of 1000.

[0045] Step 202: Generate a random number r between 0 and 1;

[0046] Step 203, if r < Then select a random action if r ≥ Then select the action with the highest current Q value.

[0047] Furthermore, a deviation value is expressed as:

[0048] ;

[0049] in, The angular deviation of the photovoltaic panel's horizontal axis is expressed as follows:

[0050] ;

[0051] The angular deviation of the photovoltaic panel's vertical axis is expressed as follows:

[0052] .

[0053] This invention provides a method for integrating heterogeneous and simultaneous transmission of photovoltaic multi-source sensors, based on the aforementioned integrated system for integrating heterogeneous and simultaneous transmission of photovoltaic multi-source sensors, and performs the following steps:

[0054] Step 301: Collect sensor data every five minutes;

[0055] Step 302: The sensor data collected every five minutes is synchronized with the time and then sent to the data fusion module;

[0056] Step 303: Perform data fusion on the sensor data using a data fusion model to obtain a fusion vector;

[0057] Step 304: Calculate the deviation between the actual angles of the horizontal and vertical axes of the photovoltaic panel and the control target angle using the fusion vector, and return to step 301.

[0058] This invention provides a device that runs the aforementioned photovoltaic multi-source sensing fusion heterogeneous simultaneous transmission integrated system through a processor.

[0059] The beneficial effects of this invention are as follows: This invention, through photovoltaic multi-source sensor fusion, can acquire and analyze various sensor data of photovoltaic panels and the environment, providing photovoltaic panel angles with high power generation efficiency and long service life. Compared with calculating deviation values ​​based on single sensor data, this system can balance multiple adjustment targets and extend the service life of photovoltaic panels. Attached Figure Description

[0060] Figure 1 This is a system module diagram of the present invention;

[0061] Figure 2 This is a flowchart of a photovoltaic multi-source sensor fusion and heterogeneous simultaneous transmission integrated method according to the present invention.

[0062] In the figure: sensor data synchronization transmission module 101, data fusion module 102, deviation detection module 103. Detailed Implementation

[0063] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0064] In a photovoltaic multi-source sensing fusion heterogeneous simultaneous transmission integrated system of the present invention, such as Figure 1 As shown, it includes:

[0065] The sensor data synchronization and transmission module 101 is used to send the sensor data to the data fusion module after time synchronization.

[0066] Sensor data is collected once at each sampling point in time. It consists of the measured values ​​detected by the sensor, including photovoltaic data and environmental data. The photovoltaic data includes the average solar irradiance of the photovoltaic panel within a preset time period, the average photovoltaic panel temperature within a preset time period, and the current horizontal and vertical axis angles of the photovoltaic panel. The environmental data includes the current solar azimuth and elevation angles, the average solar irradiance of the environment within a preset time period, the average wind speed and wind direction within a preset time period, and the average rainfall within a preset time period. The default duration of the preset time period is 5 minutes.

[0067] A time sampling point refers to the moment when data is collected at predetermined time intervals, each interval being 5 minutes.

[0068] Data fusion module 102 performs data fusion on sensor data through a data fusion model to obtain a fusion vector;

[0069] The data fusion model uses a Long Short-Term Memory (LSTM) network, which includes a forget gate, an input gate, cell states, and an output gate. The output of the forget gate... for:

[0070]

[0071] in, It is the weight matrix of the forget gate. It is the bias term of the forgetting gate. It is the hidden state of the previous time step. Sensor data at the current time step splicing, It is the Sigmoid activation function;

[0072] For the first Sensor data at each time sampling point:

[0073]

[0074] in, The average solar irradiance of the photovoltaic panels within a preset time period. The average temperature of the photovoltaic panel within a preset time period. The current horizontal axis angle of the photovoltaic panel. The current vertical axis angle of the photovoltaic panel. This is the current solar azimuth angle. This is the current solar altitude angle. The average solar radiation intensity in the environment during a preset time period. The average wind speed over a preset time period. The average wind direction over a preset time period. This represents the average rainfall over a preset time period.

[0075] Input gate output With candidate cell state The calculation method is as follows:

[0076]

[0077]

[0078] in, It is the weight matrix of the input gate. It is the bias term of the input gate. It is the hyperbolic tangent activation function. It is the weight matrix for calculating the candidate cell states. It is a bias term for the candidate cell state;

[0079] Cell state It is from the previous cell state The weighted synthesis based on candidate cell states is calculated as follows:

[0080]

[0081] in, This represents an element-wise multiplication operation;

[0082] Activation value of output gate Hidden state at the current time step The calculation method is as follows:

[0083]

[0084]

[0085] in, It is the weight matrix of the output gate. It is the bias term of the output gate;

[0086] Data fusion model outputs fusion vector , It can be represented as:

[0087]

[0088] The deviation detection module 103 calculates the deviation between the actual angles of the horizontal and vertical axes of the photovoltaic panel and the control target angle through a fusion vector.

[0089] The actual angle is In and The target angle is the action output by the decision neural network;

[0090] An action is represented as:

[0091]

[0092] in, They represent the first time. At each time sampling point, the angles that the horizontal and vertical axes of the photovoltaic panel need to be adjusted to.

[0093] A deviation value is represented as:

[0094]

[0095] in, The angular deviation of the photovoltaic panel's horizontal axis is expressed as follows:

[0096]

[0097] The angular deviation of the photovoltaic panel's vertical axis is expressed as follows:

[0098]

[0099] In one embodiment of the present invention, the first The sensor data at each time sampling point are as follows:

[0100] [980, 50, 141.7, 48.2, 146.8, 49.9, 998, 8, 32, 0]

[0101] At this moment, the horizontal axis angle of the photovoltaic panel is 141.7 degrees, the vertical axis angle is 48.2 degrees, and the decision neural network outputs the action {146.8, 49.9}. The horizontal axis angle deviation is calculated. -5.1, angular deviation of the vertical axis It is -1.7.

[0102] In one embodiment of the present invention, the calculation formula of the decision neural network is as follows:

[0103]

[0104]

[0105]

[0106]

[0107]

[0108]

[0109] in, and These represent the fusion vectors of the 1st and Lth time sampling points, respectively. , , , This represents the trainable weight parameters. , , , This represents the trainable bias parameters. Represents the dot product. This represents the Lth first intermediate feature. This represents the Lth second intermediate feature. This represents the Lth third intermediate feature. and These represent the (L-1)th and Lth output features, respectively. Let represent the output vector, where the value of the i-th component represents the Q-value of the i-th action, and tanh represents the tanh function. This represents the Sigmoid function.

[0110] The action output by the decision neural network is the action with the largest Q value.

[0111] In one embodiment of the present invention, a decision neural network is trained using Q-learning, and a reward function is set. The reward function needs to comprehensively consider the power generation efficiency, adjustment cost, environmental impact, and photovoltaic panel temperature. The power generation efficiency is the ratio of actual power generation to the maximum theoretical power generation, and the adjustment cost is the energy consumed to adjust the photovoltaic panel. The reward function is expressed as follows:

[0112]

[0113] in, This represents the immediate reward at the t-th time sampling point. This represents the actual power generation of the photovoltaic panel within five minutes prior to sampling point t. This represents the maximum theoretical power generation of the photovoltaic panel within the five minutes preceding the t time sampling points. and Let represent the deviation values ​​of the adjustment amplitude of the photovoltaic panel's horizontal and vertical axes at the t-th time sampling point, respectively. for The absolute value, for The absolute value;

[0114] This represents the environmental impact at the t-th time sampling point, assuming the photovoltaic panels were protected under harsh environmental conditions. A value of 1 indicates that the photovoltaic panel is not protected. The value is -1, otherwise The value is 0.

[0115] Harsh environment indicates the average wind speed at the t-th sampling point within a preset time period. Exceeding the wind speed threshold Average rainfall within the preset time period Exceeding the rainfall threshold ;

[0116] The criteria for determining whether to protect photovoltaic panels are:

[0117]

[0118] in This indicates the impact of adjusted wind speed on photovoltaic panels. This indicates the impact of rainfall on the photovoltaic panels after adjustment. The calculation formula is as follows:

[0119]

[0120] in, It is the angle between the wind direction and the normal direction of the photovoltaic panel;

[0121] The calculation formula is as follows:

[0122]

[0123] in, It is the angle between the direction of falling rain and the normal to the photovoltaic panel, and the calculation formula is as follows:

[0124]

[0125] in, It is the tilt angle of the photovoltaic panel relative to the horizontal plane. This is the angle at which the direction of rain falls deviates due to wind speed; the calculation formula is as follows:

[0126]

[0127] This represents the effect of photovoltaic panel temperature at the sampling point at time t. The value is the average photovoltaic panel temperature within the preset time period at the t-th time sampling point. The larger of the difference from the safe temperature threshold and 0. , , , These are weighting parameters, with default values ​​of 0.5, 0.2, 0.2, and 0.1 respectively.

[0128] In one embodiment of the present invention, the Q-learning steps include:

[0129] Step 101, replicate the decision neural network to obtain network;

[0130] Randomly select an experience from the experience pool and calculate... ;

[0131]

[0132] Representing state The maximum Q value output by the target network;

[0133] according to and Calculate the loss value and update the Q-network;

[0134]

[0135] Representing state and actions The output Q value of the lower Q network, Representing state and actions The output Q value of the target network;

[0136] This represents the state at the t-th time sampling point. This represents the action selected at the t-th time sampling point. This represents the state at the (t+1)th time sampling point;

[0137] It is a discount factor, a value between 0 and 1, used to balance the weight of immediate rewards and future rewards, with a default value of 0.6;

[0138] Step 102: Every fixed number of network updates, update the target network so that its parameters are the same as the parameters of the current Q network;

[0139] The default value for the fixed interval of network updates is 10.

[0140] Step 103 continues until the decision neural network converges or the number of iterations reaches the set number, at which point the process terminates.

[0141] The default value for iteration here is 100.

[0142] In one embodiment of the present invention, the aforementioned "done" refers to the moment when the adjustment effect is evaluated.

[0143] In one embodiment of the present invention, the method for experience acquisition is as follows: The t-th time sampling point is defined as the current time, and the action taken at the t-th time sampling point is... Execute actions The observation vector collected at the (t+1)th time sampling point thereafter Execute actions Rewards The vectors are combined into a high-dimensional vector exp and stored in the experience replay pool.

[0144] In one embodiment of the present invention, an ε-greedy strategy is used when selecting an action, and the method for selecting an action includes:

[0145] Step 201, calculate the first... Exploration parameters used to select actions during step value , The calculation method is as follows:

[0146] ;

[0147] in: At the start of training The initial value is 0.9, with a default value of 0.9. for The minimum achievable value, with a default value of 0.1; U is the total number of steps in the entire training process, with a default value of 1000.

[0148] Step 202: Generate a random number r between 0 and 1;

[0149] Step 203, if r < Then select a random action if r ≥ Then select the action with the highest current Q value.

[0150] One embodiment of the present invention provides a method for integrating heterogeneous and simultaneous transmission of photovoltaic multi-source sensors, such as... Figure 2 As shown, it includes the following steps:

[0151] Step 301: Collect sensor data every five minutes;

[0152] Step 302: The sensor data collected every five minutes is synchronized with the time and then sent to the data fusion module;

[0153] Step 303: Perform data fusion on the sensor data using a data fusion model to obtain a fusion vector;

[0154] Step 304: Calculate the deviation between the actual angles of the horizontal and vertical axes of the photovoltaic panel and the control target angle using the fusion vector, and return to step 301.

[0155] In at least one embodiment of the present invention, a storage medium is provided that stores non-transitory computer-readable instructions, which, when executed by a computer, can perform the steps of the aforementioned photovoltaic multi-source sensor fusion heterogeneous simultaneous transmission integrated method.

[0156] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A photovoltaic multi-source sensing fusion heterogeneous simultaneous transmission integrated system, characterized in that, include: The sensor data synchronization and transmission module is used to synchronize sensor data in time and then send it to the data fusion module. Sensor data is collected once at each sampling point at each time. It consists of the measured values ​​detected by the sensor, including photovoltaic data and environmental data. The photovoltaic data includes the average solar irradiance of the photovoltaic panel within a preset time period, the average photovoltaic panel temperature within a preset time period, and the current horizontal and vertical axis angles of the photovoltaic panel. The environmental data includes the current solar azimuth and elevation angles, the average solar irradiance of the environment within a preset time period, the average wind speed and wind direction within a preset time period, and the average rainfall within a preset time period. A time sampling point refers to the moment when data is collected at predetermined time intervals, each interval being 5 minutes. The data fusion module uses a data fusion model to fuse sensor data and obtain a fusion vector. The target control angle is the action output by the decision neural network. An action is represented as: ； in, They represent the first time. At each time sampling point, the angles that the horizontal and vertical axes of the photovoltaic panel need to be adjusted to; The calculation formula for a decision neural network is as follows: ； ； ； ； ； ； Among them, and These represent the fusion vectors of the 1st and Lth time sampling points, respectively. , , , This represents the trainable weight parameters. , , , This represents the trainable bias parameters. Represents the dot product. This represents the Lth first intermediate feature. This represents the Lth second intermediate feature. This represents the Lth third intermediate feature. and These represent the (L-1)th and Lth output features, respectively. Let represent the output vector, where the value of the i-th component represents the Q-value of the i-th action, and tanh represents the hyperbolic tangent activation function. This represents the Sigmoid function; the action output by the decision neural network is the action with the largest Q value. The decision neural network is trained using Q-learning. A reward function is set that comprehensively considers power generation efficiency, adjustment costs, environmental impact, and the temperature effect of the photovoltaic panels. Power generation efficiency is the ratio of actual power generation to the maximum theoretical power generation, and adjustment costs are the energy consumed to adjust the photovoltaic panels. The reward function is expressed as follows: ； in, This represents the immediate reward at the t-th time sampling point. This represents the actual power generation of the photovoltaic panel within five minutes prior to sampling point t. This represents the maximum theoretical power generation of the photovoltaic panel within the five minutes preceding the t time sampling points. and Let represent the deviation values ​​of the adjustment range of the photovoltaic panel's horizontal and vertical axes at the t-th time sampling point, respectively. This represents the environmental impact at the t-th time sampling point, assuming the photovoltaic panels were protected under harsh environmental conditions. A value of 1 indicates that the photovoltaic panel is not protected. The value is -1, otherwise The value is 0. This represents the effect of photovoltaic panel temperature at the sampling point at time t. The value is the average photovoltaic panel temperature within the preset time period at the t-th time sampling point. The maximum value between the difference from the safe temperature threshold and 0. , , , These are weighting parameters, with default values ​​of 0.5, 0.2, 0.2, and 0.1 respectively. The deviation detection module calculates the deviation between the actual angles of the horizontal and vertical axes of the photovoltaic panel and the control target angle by using fused vectors. For the first Sensor data at each time sampling point: ； in, The average solar irradiance of the photovoltaic panels within a preset time period. The average temperature of the photovoltaic panel within a preset time period. The current horizontal axis angle of the photovoltaic panel. The current vertical axis angle of the photovoltaic panel. This is the current solar azimuth angle. This is the current solar altitude angle. The average solar radiation intensity in the environment during a preset time period. The average wind speed over a preset time period. The average wind direction over a preset time period. This represents the average rainfall over a preset time period. The data fusion model uses a Long Short-Term Memory (LSTM) network, which includes a forget gate, an input gate, cell states, and an output gate. The output of the forget gate... for: ； in, It is the weight matrix of the forget gate. It is the bias term of the forgetting gate. It is the hidden state of the previous time step. Sensor data at the current time step splicing, It is the Sigmoid function; Input gate output With candidate cell state The calculation method is as follows: ； ； in, It is the weight matrix of the input gate. It is the bias term of the input gate. It is the hyperbolic tangent activation function. It is the weight matrix for calculating the candidate cell states. It is a bias term for the candidate cell state; Cell state It is from the previous cell state The weighted synthesis based on candidate cell states is calculated as follows: ； in, This represents an element-wise multiplication operation; Activation value of output gate Hidden state at the current time step The calculation method is as follows: ； ； in, It is the weight matrix of the output gate. It is the bias term of the output gate; Data fusion model outputs fusion vector , It can be represented as: ； The actual angle is In and ; The steps involved in Q learning include: Step 101, replicate the decision neural network to obtain network; Step 102: Every fixed number of network updates, update the target network so that its parameters are the same as the parameters of the current Q network; Step 103 continues until the decision neural network converges or the number of iterations reaches the set number, at which point the process terminates.

2. The photovoltaic multi-source sensor fusion heterogeneous simultaneous transmission integrated system according to claim 1, characterized in that, The ε-greedy strategy is used when selecting actions, and the methods for selecting actions include: Step 201, calculate the first... Exploration parameters used to select actions during step value , The calculation method is as follows: ； in: At the start of training The initial value is 0.9, with a default value of 0.

9. for The minimum value of , the default value is 0.1, and U is the total number of steps in the entire training process, the default value is 1000; Step 202: Generate a random number r between 0 and 1; Step 203, if r < Then select a random action if r ≥ Then select the action with the highest current Q value.

3. The photovoltaic multi-source sensor fusion heterogeneous simultaneous transmission integrated system according to claim 1, characterized in that, A deviation value is represented as: ； in, The angular deviation of the photovoltaic panel's horizontal axis is expressed as follows: ； The angular deviation of the photovoltaic panel's vertical axis is expressed as follows: 。 4. A method for integrating heterogeneous and simultaneous transmission of photovoltaic multi-source sensors, characterized in that, Based on the photovoltaic multi-source sensor fusion heterogeneous simultaneous transmission integrated system as described in any one of claims 1-3, the following steps are performed: Step 301: Collect sensor data every five minutes; Step 302: The sensor data collected every five minutes is synchronized with the time and then sent to the data fusion module; Step 303: Perform data fusion on the sensor data using a data fusion model to obtain a fusion vector; Step 304: Calculate the deviation between the actual angles of the horizontal and vertical axes of the photovoltaic panel and the control target angle using the fusion vector, and return to step 301.

5. An apparatus, characterized in that, The processor runs a photovoltaic multi-source sensor fusion heterogeneous simultaneous transmission integrated system as described in any one of claims 1-3.

Images