Burner temperature monitoring and regulating method based on deep learning

Abstract

The invention relates to the technical field of combustion engine temperature monitoring, in particular to a combustion engine temperature monitoring, regulating and controlling method based on deep learning, which comprises the following steps: acquiring multi-dimensional data of fuel characteristics and combustion states; constructing input features of the deep learning model; historical data of the combustion engine are collected, a deep learning model is built, training is carried out, and the trained deep learning model is obtained; and substituting the input features into the trained deep learning model. According to the combustion engine temperature monitoring and regulating method based on deep learning, the optimal feeding speed and the air blower power are predicted in real time by using the deep learning model through the fuel characteristics and the combustion state data obtained in real time, dynamic optimization control over the combustion process is achieved, the response time is shortened, and the combustion efficiency of the combustion engine is improved.

Description

Technical Field

[0001] This invention relates to the field of combustion engine temperature monitoring technology, specifically a method for combustion engine temperature monitoring and control based on deep learning. Background Technology

[0002] In the combustion process of a biomass burner, the burner temperature is a key parameter reflecting the combustion state, combustion efficiency, and equipment safety. The determining factors affecting the combustion temperature are the feed rate and the blower power.

[0003] Traditional burner control systems have long response times and are difficult to adapt to rapidly changing operating conditions during combustion. When abnormal temperatures occur, the system cannot make timely adjustments, often leading to incomplete combustion or overheating, which affects combustion efficiency. Therefore, a burner temperature monitoring and control method based on deep learning is proposed to solve the above problems. Summary of the Invention

[0004] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a deep learning-based method for monitoring and controlling the temperature of a burner. This method has advantages such as improving combustion efficiency and solves the problem that traditional burner control systems cannot make timely adjustments, often leading to incomplete combustion or overheating, which affects combustion efficiency.

[0005] (II) Technical Solution To achieve the aforementioned goal of improving combustion efficiency, the present invention provides the following technical solution: A method for monitoring and controlling the temperature of a combustion engine based on deep learning includes the following steps: S1. Obtain multi-dimensional data on fuel characteristics and combustion state; S2. Constructing Input Features for Deep Learning Models ; S3. Collect historical data of the burner, build a deep learning model, and train it to obtain a trained deep learning model. ; S4. Substitute the input features into the trained deep learning model to obtain the output result. ; S5. Based on the output results Analysis was conducted to draw monitoring conclusions and recommendations. .

[0006] Preferably, step S1 further includes the following sub-steps: S1.1 Real-time output of spectral matrix using a near-infrared spectrometer Through online Model predicts fuel volatiles and fixed carbon , to obtain the feature vector ; S1.2. Mass loss curves are output in real time via thermogravimetric analyzer. and heat flow signal The higher heating value was calculated. , to obtain vector : ; in, The total energy released by combustion. , , The initial mass; S1.3. Continuous distribution output via laser particle size analyzer Converted into discrete feature vectors : ; S1.4. Through Flame temperature measured by a thermocouple Temperature of the combustion completion zone ,pass and The sensors respectively obtained the flue gas and The concentration was used to calculate the combustion completeness index. , to obtain the feature vector : ; in, The ideal value is approximately 5%. As a safety threshold, These are the weighting coefficients.

[0007] Preferably, step S2 further includes the following sub-steps: S2.1. Concatenate the feature vectors to obtain the concatenated vector. : ; S2.2. For each feature vector Calculate its weighting effect on complete combustion. : ; in, for variance For conditional expectation; S2.3. Retain Based on the features, construct the final input features: ; S2.4. Obtain the final input feature vector: .

[0008] Preferably, step S3 includes the following sub-steps: S3.1.1. Collection Historical combustion data for each group of burners, including input features for each group of historical combustion data. [ , and output target ; S3.1.2. Perform feature standardization, using the following formula: ; in, The feature mean vector, is the characteristic standard deviation vector.

[0009] Preferably, step S3 further includes the following sub-steps: S3.2.1. Constructing a fully connected layer neural network model : ; in, For the standardized input features, This is the first layer weight matrix. This is the first layer bias vector. This is the weight matrix for the second layer. This is the second layer bias vector. For activation functions; S3.2.2. Design the mean square error loss function : ; in, The number of training samples. For the first The true target value of each sample For the first The predicted target value for each sample; S3.2.3. Define the output layer gradient and the hidden layer gradient: ; ; ; ; in, This is an indicator function; it returns 1 if the condition is true, and 0 otherwise. S3.2.4. Define parameter update rules; S3.2.5. Construct a fully connected layer neural network model Verification and evaluation.

[0010] Preferably, step S4 further includes the following sub-steps: S4.1. Process the input features Standardization process: ; S4.2. Standardize the features Input into the trained fully connected neural network model In the middle, the output is obtained : .

[0011] Preferably, step S5 further includes the following sub-steps: S5.1.1. Obtain the feed rate of the feed valve. and the power of the blower ; S5.1.2. Calculate the feed rate error Error with blower power : ; ; S5.1.3. Based on feed rate error Error with blower power Obtain evaluation factors : .

[0012] Preferably, step S5 further includes the following sub-steps: S5.2.1. Based on evaluation factors Make a judgment when or hour, ; S5.2.2. When or hour, “ The effect is mediocre and adjustments are needed. The feed rate of the feed valve should be adjusted to... The power of the blower is adjusted to ”; S5.2.3. hour, “ The combustion effect is poor, and the feed valve or blower needs to be inspected and repaired.

[0013] (III) Beneficial Effects Compared with existing technologies, this invention provides a method for monitoring and controlling the temperature of a burner based on deep learning, which has the following advantages: This deep learning-based method for monitoring and controlling burner temperature uses real-time acquired fuel characteristics and combustion state data to predict the optimal feed rate and blower power in real time, achieving dynamic optimization control of the combustion process, shortening the response time, and improving the combustion efficiency of the burner. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of a deep learning-based method for monitoring and controlling the temperature of a burner proposed in this invention. Detailed Implementation

[0015] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] A method for monitoring and controlling the temperature of a combustion engine based on deep learning includes the following steps: S1. Obtain multi-dimensional data on fuel characteristics and combustion state; S2. Constructing Input Features for Deep Learning Models ; S3. Collect historical data of the burner, build a deep learning model, and train it to obtain a trained deep learning model. ; S4. Substitute the input features into the trained deep learning model to obtain the output result. ; S5. Based on the output results Analysis was conducted to draw monitoring conclusions and recommendations. .

[0017] In this embodiment, step S1 further includes the following sub-steps: S1.1 Real-time output of spectral matrix using a near-infrared spectrometer Through online Model predicts fuel volatiles and fixed carbon , to obtain the feature vector ; via online Model predicts fuel volatiles and fixed carbon The specific steps are as follows: set up The model is : in, For bias, For the target variable, The bias matrix, , Number of wavelength points The number of samples; By minimizing the loss function : ; Standardize the samples. Then we get: ; S1.2. Mass loss curves are output in real time via thermogravimetric analyzer. and heat flow signal The higher heating value was calculated. , to obtain vector : ; in, The total energy released by combustion. , , The initial mass; S1.3. Continuous distribution output via laser particle size analyzer Converted into discrete feature vectors : ; Feature vector The derivation process: Define particle size range (like ); The quality score for each interval is: ; Fitting the data using the Rossin-Lammler distribution ,get : .

[0018] S1.4. Through Flame temperature measured by a thermocouple Temperature of the combustion completion zone ,pass and The sensors respectively obtained the flue gas and The concentration was used to calculate the combustion completeness index. , to obtain the feature vector : ; in, The ideal value is approximately 5%. As a safety threshold, These are the weighting coefficients.

[0019] In this embodiment, step S2 further includes the following sub-steps: S2.1. Concatenate the feature vectors to obtain the concatenated vector. : ; S2.2. For each feature vector Calculate its weighting effect on complete combustion. : ; in, for variance For conditional expectation; S2.3. Retain Based on the features, construct the final input features: ; S2.4. Obtain the final input feature vector: .

[0020] In this embodiment, step S3 includes the following sub-steps: S3.1.1. Collection Historical combustion data for each group of burners, including input features for each group of historical combustion data. [ , and output target ; S3.1.2. Perform feature standardization, using the following formula: ; in, The feature mean vector, is the characteristic standard deviation vector.

[0021] In this embodiment, step S3 further includes the following sub-steps: S3.2.1. Constructing a fully connected layer neural network model : ; in, For the standardized input features, This is the first layer weight matrix. This is the first layer bias vector. This is the weight matrix for the second layer. This is the second layer bias vector. For activation functions; Number of hidden layer neurons The number of neurons in the output layer is 2.

[0022] S3.2.2. Design the mean square error loss function : ; in, The number of training samples. For the first The true target value of each sample For the first The predicted target value for each sample; S3.2.3. Define the output layer gradient and the hidden layer gradient: ; ; ; ; in, This is an indicator function; it returns 1 if the condition is true, and 0 otherwise. S3.2.4. Define parameter update rules; ; ; ; ; in, The learning rate (usually taken as...) ).

[0023] S3.2.5. Construct a fully connected layer neural network model Verification and evaluation.

[0024] The dataset was divided into 70% training set. 15% validation set and 15% of the test set ; Calculate the mean square error loss of the validation set ; Calculate the improvement rate of combustion completeness .

[0025] In this embodiment, step S4 further includes the following sub-steps: S4.1. Process the input features Standardization process: ; S4.2. Standardize the features Input into the trained fully connected neural network model In the middle, the output is obtained : .

[0026] In this embodiment, step S5 further includes the following sub-steps: S5.1.1. Obtain the feed rate of the feed valve. and the power of the blower ; S5.1.2. Calculate the feed rate error Error with blower power : ; ; S5.1.3. Based on feed rate error Error with blower power Obtain evaluation factors : .

[0027] In this embodiment, step S5 further includes the following sub-steps: S5.2.1. Based on evaluation factors Make a judgment when or hour, ; S5.2.2. When or hour, “ The effect is mediocre and adjustments are needed. The feed rate of the feed valve should be adjusted to... The power of the blower is adjusted to ”; S5.2.3. hour, “ The combustion effect is poor, and the feed valve or blower needs to be inspected and repaired.

[0028] The beneficial effects of this invention are: the deep learning-based burner temperature monitoring and control method uses real-time acquired fuel characteristics and combustion state data, and a deep learning model to predict the optimal feed rate and blower power in real time, thereby achieving dynamic optimization control of the combustion process, shortening the response time, and improving the combustion efficiency of the burner.

[0029] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring and controlling the temperature of a combustion engine based on deep learning, characterized in that, Includes the following steps: S1. Obtain multi-dimensional data on fuel characteristics and combustion state; S2. Constructing Input Features for Deep Learning Models ; S3. Collect historical data of the burner, build a deep learning model, and train it to obtain a trained deep learning model. ; S4. Substitute the input features into the trained deep learning model to obtain the output result. ; S5. Based on the output results Analysis was conducted to draw monitoring conclusions and recommendations. .

2. The method for monitoring and controlling combustion engine temperature based on deep learning according to claim 1, characterized in that, Step S1 also includes the following sub-steps: S1.1 Real-time output of spectral matrix using a near-infrared spectrometer Through online Model predicts fuel volatiles and fixed carbon , to obtain the feature vector ; S1.

2. Mass loss curves are output in real time via thermogravimetric analyzer. and heat flow signal The higher heating value was calculated. , to obtain vector : ； in, The total energy released by combustion. , , The initial mass; S1.

3. Continuous distribution output via laser particle size analyzer Converted into discrete feature vectors : ； S1.

4. Through Flame temperature measured by a thermocouple Temperature of the combustion completion zone ,pass and The sensors respectively obtained the flue gas and The concentration was used to calculate the combustion completeness index. , to obtain the feature vector : ； in, The ideal value is approximately 5%. As a safety threshold, These are the weighting coefficients.

3. The method for monitoring and controlling combustion engine temperature based on deep learning according to claim 2, characterized in that, Step S2 also includes the following sub-steps: S2.

1. Concatenate the feature vectors to obtain the concatenated vector. : ； S2.

2. For each feature vector Calculate its weighting effect on complete combustion. : ； in, for variance For conditional expectation; S2.

3. Retain Based on the features, construct the final input features: ； S2.

4. Obtain the final input feature vector: 。 4. The method for monitoring and controlling combustion engine temperature based on deep learning according to claim 1, characterized in that, Step S3 includes the following sub-steps: S3.1.

1. Collection Historical combustion data for each group of burners, including input features for each group of historical combustion data. [ , and output target ; S3.1.

2. Perform feature standardization, using the following formula: ； in, The feature mean vector, is the feature standard deviation vector.

5. The method for monitoring and controlling combustion engine temperature based on deep learning according to claim 1, characterized in that, Step S3 also includes the following sub-steps: S3.2.

1. Constructing a fully connected neural network model : ； in, For the standardized input features, This is the first layer weight matrix. This is the first layer bias vector. This is the weight matrix for the second layer. This is the second layer bias vector. For activation functions; S3.2.

2. Design the mean square error loss function : ； in, The number of training samples. For the first The true target value of each sample For the first The predicted target value for each sample; S3.2.

3. Define the output layer gradient and the hidden layer gradient: ； ； ； ； in, This is an indicator function; it returns 1 if the condition is true, and 0 otherwise. S3.2.

4. Define parameter update rules; S3.2.

5. Construct a fully connected layer neural network model Verification and evaluation.

6. The method for monitoring and controlling the temperature of a combustion engine based on deep learning according to claim 5, characterized in that, Step S4 also includes the following sub-steps: S4.

1. Process the input features Standardization process: ； S4.

2. Standardize the features Input into the trained fully connected neural network model In the middle, the output is obtained : 。 7. The method for monitoring and controlling the temperature of a combustion engine based on deep learning according to claim 1, characterized in that, Step S5 also includes the following sub-steps: S5.1.

1. Obtain the feed rate of the feed valve. and the power of the blower ; S5.1.

2. Calculate the feed rate error Error with blower power : ； ； S5.1.

3. Based on feed rate error Error with blower power Obtain evaluation factors : 。 8. The method for monitoring and controlling the temperature of a combustion engine based on deep learning according to claim 7, characterized in that, Step S5 also includes the following sub-steps: S5.2.

1. Based on evaluation factors Make a judgment when or hour, ; S5.2.

2. When or hour, " The effect is mediocre and adjustments are needed. The feed rate of the feed valve should be adjusted to... The power of the blower is adjusted to ”; S5.2.

3. hour, " The combustion effect is poor, and the feed valve or blower needs to be inspected and repaired.

Images