Valve identification method and device based on neural network, medium and equipment

Abstract

The invention discloses a valve identification method and device based on a neural network, a medium and equipment. The method comprises the following steps: acquiring voltage data, current data, flow data and valve opening data in a continuous time period; inputting the voltage data, the current data and the flow data at each moment in the continuous time period into a pre-trained neural network model to obtain a load at each moment; obtaining a load curve based on the load at each moment and the valve opening data at each moment in the continuous time period; and respectively calculating the average loss of the load curve and the standard load curves of different types of valves, and outputting the type of the valve corresponding to the standard load curve when the average loss is minimum. The valve type can be recognized, manual checking of the valve type can be avoided, and the labor cost can be saved.

Description

Technical Field

[0001] The invention relates to a valve identification method, device, medium and equipment based on neural network, belonging to the technical field of valve control. Background Art

[0002] As one of the most widely used machines in the field of industrial control, valves have now become key equipment for controlling pipeline flow.

[0003] The effect of valve opening control by the valve control device directly affects the flow control effect of the entire pipeline network, and thus affects the overall efficiency of the factory.

[0004] The valve control device needs to model the valve, and the load models of different types of valves are different. In thermal power plants, the high-temperature piping system environment is relatively harsh, and manual inspection of valve types is costly and risky. Summary of the invention

[0005] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a valve identification method, device, medium and equipment based on a neural network, which can identify the valve type, avoid manual checking of the valve type, and save labor costs. To achieve the above purpose, the present invention is implemented by adopting the following technical solutions:

[0006] In a first aspect, the present invention provides a valve identification method based on a neural network, comprising:

[0007] Obtain voltage data, current data, flow data and valve opening data within a continuous time period;

[0008] Input the voltage data, current data and flow data at each moment in a continuous time period into a pre-trained neural network model to obtain the load at each moment;

[0009] Based on the load at each moment and the valve opening data at each moment in a continuous time period, a load curve is obtained;

[0010] Calculate the average loss of the load curve and the standard load curve of different types of valves respectively, and output the type of valve corresponding to the standard load curve when the average loss is minimum.

[0011] In combination with the first aspect, optionally, inputting the voltage data, current data and flow data at each moment in the continuous time period into a pre-trained neural network model to obtain the load at each moment includes:

[0012] The voltage data, current data and flow data acquired in a continuous time period are smoothed and pre-processed to eliminate invalid data;

[0013] The voltage data, current data and flow data after smoothing preprocessing are grouped according to each moment to obtain the voltage data, current data and flow data at each moment;

[0014] The voltage data, current data and flow data at each moment are used as the input vector of the pre-trained neural network model, and the output vector obtained is the load at each moment.

[0015] In combination with the first aspect, optionally, the pre-trained neural network model is trained based on sample data using an error back propagation method, and is trained by the following steps:

[0016] Step a: Initialize the weights and biases of the pre-built neural network model;

[0017] Step b: Input the sample value of the sample data into the neural network model, perform forward propagation calculation, and obtain the output value;

[0018] Step c: Calculate the loss function between each output value and the corresponding sample output value, and use the gradient descent algorithm to update the weights and biases of the neural network model based on the loss function to obtain an updated neural network model;

[0019] Step d: Repeat the iterative calculations of step b and step c, and in response to the loss function between each output value and the sample value corresponding to the output value being less than a preset loss threshold or the number of iterations being equal to a preset number threshold, obtain a pre-trained neural network model.

[0020] In combination with the first aspect, optionally, the forward propagation calculation is expressed by the following formula:

[0021] ;

[0022] in, x i is the sample value of the sample data, and the input vector is expressed as , m is the total number of sample data; is the weight from the input layer to the hidden layer, and the weight matrix from the input layer to the hidden layer is expressed as ; is the bias from the input layer to the hidden layer; is the activation function from the input layer to the hidden layer;

[0023] in, h i is the hidden layer node value, and the hidden layer vector is expressed as , p is the number of hidden layer nodes; the output vector composed of output values ​​is expressed as , s is the total number of output values; is the weight from the hidden layer to the output layer, and the weight matrix from the hidden layer to the output layer is expressed as ; is the bias from the hidden layer to the output layer; is the activation function from the hidden layer to the output layer.

[0024] In combination with the first aspect, optionally, the loss function for calculating each output value and the sample value corresponding to the output value is to calculate a mean square error, which is expressed by the following formula:

[0025] ;

[0026] in, is the output value, and the vector that makes up the output is expressed as , s is the total number of output values; is the sample output value, and the vector is expressed as , s is the total number of output values; It is the loss function between the calculated output value and the corresponding sample output value.

[0027] In combination with the first aspect, optionally, the weights and biases of the neural network model are updated using a gradient descent algorithm based on the loss function, which is expressed by the following formula:

[0028] ;

[0029] in, is the weight matrix of the neural network model composed of the weight matrix from the input layer to the hidden layer and the weight matrix from the hidden layer to the output layer. is the weight matrix obtained for each update; is the previous weight matrix, is the previous bias; α is the step size of each update, is the bias of the neural network model consisting of the bias from the input layer to the hidden layer and the bias from the hidden layer to the output layer. is the bias obtained for each update.

[0030] In combination with the first aspect, optionally, respectively calculating the average loss of the load curve and the standard load curves of different types of valves includes:

[0031] The horizontal coordinate of the load curve is the valve opening, and the vertical coordinate is the load;

[0032] When the valve opening is x, calculate the loss xn between the actual load value on the load curve and the standard load value on the standard load curve of the nth type of valve; where the value range of x is [0,1], and n is a positive integer;

[0033] Adjust the value of x to obtain multiple sets of losses xn;

[0034] Calculate the mean of multiple sets of losses xn to obtain the average loss An between the load curve and the standard load curve of the nth type of valve;

[0035] Compare the size of the average loss An when n is different values, and find the type of valve corresponding to the standard load curve when the output average loss is minimum.

[0036] In a second aspect, the present invention provides a valve identification device based on a neural network, comprising:

[0037] An acquisition module, used to acquire voltage data, current data, flow data and valve opening data within a continuous time period;

[0038] The load output module is used to input the voltage data, current data and flow data at each moment in a continuous time period into a pre-trained neural network model to obtain the load at each moment;

[0039] A load curve output module is used to obtain a load curve based on the load at each moment and the valve opening data at each moment in a continuous time period;

[0040] The valve type output module is used to calculate the average loss of the load curve and the standard load curve of different types of valves, and output the type of valve corresponding to the standard load curve when the average loss is minimum.

[0041] In a third aspect, the present invention provides a computer-readable storage medium having a computer program / instruction stored thereon, which, when executed by a processor, implements the steps of the neural network-based valve identification method described in the first aspect.

[0042] In a fourth aspect, the present invention provides a computer device, comprising:

[0043] Memory, for storing computer programs / instructions;

[0044] A processor is used to execute the computer program / instructions to implement the steps of the neural network-based valve identification method described in the first aspect.

[0045] Compared with the prior art, the valve identification method, device, medium and equipment based on neural network provided by the embodiment of the present invention have the following beneficial effects:

[0046] The present invention obtains voltage data, current data, flow data and valve opening data within a continuous time period; inputs the voltage data, current data and flow data at each moment within the continuous time period into a pre-trained neural network model to obtain the load at each moment; obtains a load curve based on the load at each moment and the valve opening data at each moment within the continuous time period; the present invention uses a sampling circuit in conjunction with a computer to obtain voltage data, current data, flow data and valve opening data, the circuit structure is simple and easy to implement industrially, no on-site manual sampling is required, and labor costs are saved; the present invention uses a neural network to obtain the load at each moment, and compared with the existing small squares load observer or extended Kalman filter algorithm to identify torque, the present invention will not be affected by the working environment, has strong anti-disturbance ability, and can accurately obtain the load and load curve;

[0047] The present invention respectively calculates the average loss of the load curve and the standard load curve of different types of valves, and outputs the type of valve corresponding to the standard load curve when the average loss is minimum; the present invention can perform offline identification, which greatly reduces the computing resources required on the chip; the present invention can identify the valve type, can avoid manual checking of the valve type, can save labor costs, and after obtaining the identification result, it is convenient to call the load model corresponding to the valve type. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 1 is a flow chart of a valve identification method based on a neural network provided in Example 1 of the present invention;

[0049] Figure 2 Schematic diagram of the overall architecture of a valve identification method based on a neural network provided in Example 1 of the present invention;

[0050] Figure 3 It is a structural schematic diagram of a neural network in a valve identification method based on a neural network provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0051] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0052] Embodiment 1:

[0053] like Figure 1 As shown, the present invention provides a valve identification method based on a neural network, comprising:

[0054] Obtain voltage data, current data, flow data and valve opening data within a continuous time period;

[0055] Input the voltage data, current data and flow data at each moment in a continuous time period into a pre-trained neural network model to obtain the load at each moment;

[0056] Based on the load at each moment and the valve opening data at each moment in a continuous time period, a load curve is obtained;

[0057] Calculate the average loss of the load curve and the standard load curve of different types of valves respectively, and output the type of valve corresponding to the standard load curve when the average loss is minimum.

[0058] The specific steps include:

[0059] Step 1: Obtain voltage data, current data, flow data and valve opening data within a continuous time period.

[0060] like Figure 2 As shown, the circuit of this embodiment is provided with an asynchronous motor, a drive circuit, a power circuit, a sampling circuit, a digital processing control module and an auxiliary power supply circuit. Among them, the power circuit includes a three-phase rectifier bridge and an IGBT power module, and the sampling circuit includes a bus voltage sampling circuit, a phase current sampling circuit, and an overcurrent protection circuit sampling signal.

[0061] Specifically, the digital processing module is a DSC chip, model ADP16F01.

[0062] The input end of the drive circuit is connected to the digital processing control module via ePWM, and the output end of the drive circuit is connected to the three wiring terminals of the asynchronous motor via the IGBT power module of the power circuit to provide a drive signal for the asynchronous motor.

[0063] The input end of the three-phase rectifier bridge of the power circuit is connected to a 380V AC power supply via a rectifier filter circuit, and the output end of the three-phase rectifier bridge of the power circuit is connected to a three-phase power line of an asynchronous motor.

[0064] The bus voltage sampling circuit is connected between the output end of the three-phase rectifier bridge and the three-phase power line of the asynchronous motor, and the phase current sampling circuit is connected to the three-phase power line of the asynchronous motor. The overcurrent protection circuit sampling signal is connected to the output end of the three-phase rectifier bridge.

[0065] The output signals of the bus voltage sampling circuit, the phase current sampling circuit and the overcurrent protection circuit sampling signals are input to the ADC port of the digital processing control module and transmitted to the computer by the communication circuit of the digital processing control module.

[0066] The valve is provided with a flow sensor and a position sensor. The flow sensor is used to output flow data, and the position sensor is used to collect the valve opening position and output valve opening data. The output signals of the flow sensor and the position sensor are input into the ADC port of the digital processing control module and transmitted to the computer by the communication circuit of the digital processing control module.

[0067] Specifically, the communication circuit is an RS422 serial communication circuit.

[0068] The auxiliary power supply circuit is composed of a power chip, a starting circuit, a transformer, a multi-channel output port, and an optocoupler isolation sampling circuit. The auxiliary power supply circuit is used to power the drive circuit, the sampling circuit, and the digital processing control module.

[0069] Specifically, the power chip of the auxiliary power supply circuit is a CSV35022 low-side control chip.

[0070] This embodiment uses a sampling circuit in conjunction with a computer to obtain voltage data, current data, flow data and valve opening data. The circuit structure is simple and easy to implement industrially. There is no need for on-site manual sampling, which saves labor costs.

[0071] Step 2: Input the voltage data, current data and flow data at each moment in the continuous time period into the pre-trained neural network model to obtain the load at each moment.

[0072] Step 2.1: Smooth the voltage data, current data, and flow data acquired in a continuous time period and pre-process them to eliminate invalid data.

[0073] Step 2.2: Group the voltage data, current data, and flow data after smoothing preprocessing according to each moment to obtain the voltage data, current data, and flow data at each moment.

[0074] Step 2.3: The voltage data, current data and flow data at each moment are used as the input vector of the pre-trained neural network model, and the output vector obtained is the load at each moment.

[0075] The pre-trained neural network model is trained based on sample data using the error back propagation method.

[0076] The error back propagation method is divided into two stages: signal forward propagation and error back propagation. In the forward propagation stage, the sample signal enters the neural network from the input side, and is transmitted to the output layer after calculation layer by layer. In the error back propagation stage, according to the output value and the corresponding loss function, it is propagated to the input side layer by layer, and then enters the forward propagation stage again, and iterates in this cycle until the output result meets the requirements or reaches the specified number of iterations.

[0077] The training is carried out through the following steps:

[0078] Step a: Initialize the weights and biases of the pre-built neural network model.

[0079] Step b: Input the sample value of the sample data into the neural network model, perform forward propagation calculation, and obtain the output value.

[0080] In this embodiment, forward propagation calculation is performed, which is expressed by the following formula:

[0081] ;

[0082] in, x i is the sample value of the sample data, and the input vector is expressed as , m is the total number of sample data; is the weight from the input layer to the hidden layer, and the weight matrix from the input layer to the hidden layer is expressed as ; is the bias from the input layer to the hidden layer; is the activation function from the input layer to the hidden layer;

[0083] in, h i is the hidden layer node value, and the hidden layer vector is expressed as , p is the number of hidden layer nodes; the output vector composed of output values ​​is expressed as , s is the total number of output values; is the weight from the hidden layer to the output layer, and the weight matrix from the hidden layer to the output layer is expressed as ; is the bias from the hidden layer to the output layer; is the activation function from the hidden layer to the output layer.

[0084] Step c: Calculate the loss function between each output value and the corresponding sample output value, and use the gradient descent algorithm based on the loss function to update the weights and biases of the neural network model to obtain an updated neural network model.

[0085] In this embodiment, the loss function between each output value and the sample value corresponding to the output value is to calculate the mean square error, which is expressed by the following formula:

[0086] ;

[0087] in, is the output value, and the vector that makes up the output is expressed as , s is the total number of output values; is the sample output value, and the vector is expressed as , s is the total number of output values; It is the loss function between the calculated output value and the corresponding sample output value.

[0088] The weights and biases of the neural network model are updated using the gradient descent algorithm based on the loss function, which is expressed as follows:

[0089] ;

[0090] in, is the weight matrix of the neural network model composed of the weight matrix from the input layer to the hidden layer and the weight matrix from the hidden layer to the output layer. is the weight matrix obtained for each update; is the previous weight matrix, is the previous bias; α is the step size of each update, is the bias of the neural network model consisting of the bias from the input layer to the hidden layer and the bias from the hidden layer to the output layer. is the bias obtained for each update.

[0091] It should be noted that the step length α The larger it is, the faster the model converges, but it may cause the model to oscillate.

[0092] Step d: Repeat the iterative calculations of step b and step c, and in response to the loss function between each output value and the sample value corresponding to the output value being less than a preset loss threshold or the number of iterations being equal to a preset number threshold, obtain a pre-trained neural network model.

[0093] It should be noted that the number of input nodes, output nodes, and hidden layer nodes need to be adjusted according to the results of training in order to achieve the best effect of the neural network model.

[0094] like Figure 3 As shown, in this embodiment, the number of input layer nodes of the neural network model is 64, the number of output nodes is 3, the number of hidden layers is 1, and the number of nodes in each hidden layer is 128.

[0095] In this embodiment, the activation function is ReLu, and the regularization masking rate is set to 0.2.

[0096] This embodiment uses a neural network to obtain the load at each moment. Compared with the existing small squares load observer or extended Kalman filter algorithm to identify torque, it is not affected by the working environment, has strong anti-disturbance ability, and can accurately obtain the load.

[0097] Step 3: Based on the load at each moment and the valve opening data at each moment in the continuous time period, a load curve is obtained.

[0098] The horizontal axis of the load curve is the valve opening, and the vertical axis is the load.

[0099] Step 4: Calculate the average loss of the load curve and the standard load curve of different types of valves respectively, and output the type of valve corresponding to the standard load curve when the average loss is minimized.

[0100] Step 4.1: When the valve opening is x, calculate the loss xn between the actual load value on the load curve and the standard load value on the standard load curve of the nth type of valve.

[0101] The value range of x is [0,1], and n is a positive integer.

[0102] Step 4.2: Adjust the value of x to obtain multiple sets of losses xn.

[0103] Step 4.3: Calculate the mean of multiple sets of losses xn to obtain the average loss An of the load curve and the standard load curve of the nth type of valve.

[0104] Step 4.4: Compare the average loss An when n is different values, and find the valve type corresponding to the standard load curve when the output average loss is minimum.

[0105] This embodiment can perform offline identification, greatly reducing the computing resources required on the chip; the present invention can identify the valve type, can avoid manual checking of the valve type, can save labor costs, and after obtaining the identification result, it is convenient to call the load model corresponding to the valve type.

[0106] Embodiment 2:

[0107] The present invention provides a valve identification device based on a neural network, comprising:

[0108] An acquisition module, used to acquire voltage data, current data, flow data and valve opening data within a continuous time period;

[0109] The load output module is used to input the voltage data, current data and flow data at each moment in a continuous time period into a pre-trained neural network model to obtain the load at each moment;

[0110] A load curve output module is used to obtain a load curve based on the load at each moment and the valve opening data at each moment in a continuous time period;

[0111] The valve type output module is used to calculate the average loss of the load curve and the standard load curve of different types of valves, and output the type of valve corresponding to the standard load curve when the average loss is minimum.

[0112] Embodiment 3:

[0113] The present invention provides a computer-readable storage medium having a computer program / instruction stored thereon. When the computer program / instruction is executed by a processor, the steps of the valve identification method based on a neural network described in Example 1 are implemented.

[0114] Embodiment 4:

[0115] The present invention provides a computer device, comprising:

[0116] Memory, for storing computer programs / instructions;

[0117] A processor is used to execute the computer program / instructions to implement the steps of the neural network-based valve identification method described in Example 1.

[0118] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0119] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0120] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0121] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0122] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which all fall within the protection of the present invention.

Claims

1. A valve identification method based on neural network, characterized in that: include: Obtain voltage data, current data, flow data and valve opening data within a continuous time period; Input the voltage data, current data and flow data at each moment in a continuous time period into a pre-trained neural network model to obtain the load at each moment; Based on the load at each moment and the valve opening data at each moment in a continuous time period, a load curve is obtained; Calculate the average loss of the load curve and the standard load curve of different types of valves respectively, and output the type of valve corresponding to the standard load curve when the average loss is minimum.

2. The valve identification method based on neural network according to claim 1, characterized in that: The voltage data, current data and flow data at each moment in the continuous time period are input into the pre-trained neural network model to obtain the load at each moment, including: The voltage data, current data and flow data acquired in a continuous time period are smoothed and pre-processed to eliminate invalid data; The voltage data, current data and flow data after smoothing preprocessing are grouped according to each moment to obtain the voltage data, current data and flow data at each moment; The voltage data, current data and flow data at each moment are used as the input vector of the pre-trained neural network model, and the output vector obtained is the load at each moment.

3. The valve identification method based on neural network according to claim 1, characterized in that: The pre-trained neural network model is trained based on sample data using an error back propagation method, and is trained by the following steps: Step a: Initialize the weights and biases of the pre-built neural network model; Step b: Input the sample value of the sample data into the neural network model, perform forward propagation calculation, and obtain the output value; Step c: Calculate the loss function between each output value and the corresponding sample output value, and use the gradient descent algorithm to update the weights and biases of the neural network model based on the loss function to obtain an updated neural network model; Step d: Repeat the iterative calculations of step b and step c, and in response to the loss function between each output value and the sample value corresponding to the output value being less than a preset loss threshold or the number of iterations being equal to a preset number threshold, obtain a pre-trained neural network model.

4. The valve identification method based on neural network according to claim 3 is characterized in that: The forward propagation calculation is expressed by the following formula: ； in, x i is the sample value of the sample data, and the input vector is expressed as , m is the total number of sample data; is the weight from the input layer to the hidden layer, and the weight matrix from the input layer to the hidden layer is expressed as ; is the bias from the input layer to the hidden layer; is the activation function from the input layer to the hidden layer; in, h i is the hidden layer node value, and the hidden layer vector is expressed as , p is the number of hidden layer nodes; the output vector composed of output values ​​is expressed as , s is the total number of output values; is the weight from the hidden layer to the output layer, and the weight matrix from the hidden layer to the output layer is expressed as ; is the bias from the hidden layer to the output layer; is the activation function from the hidden layer to the output layer.

5. The valve identification method based on neural network according to claim 3 is characterized in that: The loss function for calculating the loss between each output value and the sample value corresponding to the output value is to calculate the mean square error, which is expressed by the following formula: ； in, is the output value, and the vector that makes up the output is expressed as , s is the total number of output values; is the sample output value, and the vector is expressed as , s is the total number of output values; It is the loss function between the calculated output value and the corresponding sample output value.

6. The valve identification method based on neural network according to claim 5, characterized in that: The weights and biases of the neural network model are updated using the gradient descent algorithm based on the loss function, which is expressed by the following formula: ； in, is the weight matrix of the neural network model composed of the weight matrix from the input layer to the hidden layer and the weight matrix from the hidden layer to the output layer. is the weight matrix obtained for each update; is the previous weight matrix, is the previous bias; α is the step size of each update, is the bias of the neural network model consisting of the bias from the input layer to the hidden layer and the bias from the hidden layer to the output layer. is the bias obtained for each update.

7. The valve identification method based on neural network according to claim 1, characterized in that: The calculation of the average loss of the load curve and the standard load curve of different types of valves respectively includes: The horizontal coordinate of the load curve is the valve opening, and the vertical coordinate is the load; When the valve opening is x, calculate the loss xn between the actual load value on the load curve and the standard load value on the standard load curve of the nth type of valve; where the value range of x is [0,1], and n is a positive integer; Adjust the value of x to obtain multiple sets of losses xn; Calculate the mean of multiple sets of losses xn to obtain the average loss An between the load curve and the standard load curve of the nth type of valve; Compare the size of the average loss An when n is different values, and find the type of valve corresponding to the standard load curve when the output average loss is minimum.

8. A valve identification device based on neural network, characterized in that: include: An acquisition module, used to acquire voltage data, current data, flow data and valve opening data within a continuous time period; The load output module is used to input the voltage data, current data and flow data at each moment in a continuous time period into a pre-trained neural network model to obtain the load at each moment; A load curve output module is used to obtain a load curve based on the load at each moment and the valve opening data at each moment in a continuous time period; The valve type output module is used to calculate the average loss of the load curve and the standard load curve of different types of valves, and output the type of valve corresponding to the standard load curve when the average loss is minimum.

9. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instruction is executed by a processor, the steps of the valve identification method based on a neural network described in any one of claims 1-7 are implemented.

10. A computer device, characterized in that: include: Memory, for storing computer programs / instructions; A processor, configured to execute the computer program / instructions to implement the steps of the neural network-based valve identification method according to any one of claims 1 to 7.

Images