Compressor operation load prediction method and device, electronic product and medium

Abstract

The invention provides a compressor operation load prediction method and device, an electronic product and a medium. The method comprises the following steps: acquiring historical data from a database as a training set; performing normalization processing on the historical data; constructing a 1DCNN network model according to an actual problem; inputting the data in the training set into the 1DCNN network model to obtain a real-time predicted value of the 1DCNN network model to the operation load variable, and optimizing the network structure and parameters according to the difference between the real-time predicted value and the actual value; and applying the optimized 1DCNN network model to online prediction of operation load variables. According to the compressor operation load prediction method and device, the electronic product and the medium, an effective solution is provided for modeling in the compressor assembly process.

Description

Technical Field

[0001] The present invention belongs to the field of neural networks, and in particular, relates to a method, device, electronic product and medium for predicting the operation load of a compressor. Background Art

[0002] The compressor is one of the core components of the refrigeration system, and its operating status directly affects the performance and energy efficiency of the refrigeration system. During the compressor assembly process, it is crucial to ensure the correct assembly of each component and accurate prediction of operating parameters. Assembly data provides important information about the compressor manufacturing process and performance characteristics. Accurate assembly ensures the final performance of the compressor product. The traditional assembly process relies on experience and trial and error, and it is difficult to make dynamic adjustments under different working conditions, which leads to problems such as low assembly efficiency and unstable assembly quality.

[0003] The compressor assembly process involves numerous assembly variables. The multivariable complexity of the process makes it difficult to manually monitor and adjust all relevant parameters, increasing the complexity of the assembly process. At the same time, a large amount of data needs to be monitored and analyzed, but traditional assembly methods rely on experience and it is difficult to efficiently process and interpret this data to support real-time control and improvement of assembly quality.

[0004] Deep learning technology, especially convolutional neural network (CNN), has achieved remarkable results in image recognition, natural language processing and other fields. One-dimensional convolutional neural network (1DCNN) is a deep learning model for sequence data, which can be used for data feature extraction and prediction. Therefore, applying 1DCNN to the prediction of compressor load parameters is a potential method.

[0005] In response to the above challenges, the present invention proposes a compressor operation load prediction model based on 1DCNN for the compressor assembly process, which can effectively automatically judge the compressor operation status through the load data, thereby judging whether there are any problems in the compressor assembly and monitoring the assembly process in real time. It provides core technology and system solutions for the automated quality inspection of the compressor assembly process.

[0006] The compressor assembly site environment is highly complex, and the assembly process may involve multiple complex steps and parameters, which need to be monitored in real time to ensure the quality and performance of the assembly. Usually, the assembly data reflects the key performance parameters and assembly quality of the compressor, which directly affect the operating performance of the compressor. For example, parameters such as cylinder height, piston flatness, and blade height affect the internal structure and air flow characteristics of the compressor, thereby affecting key performance indicators such as current, flow, power, magnetic flux, and temperature. Therefore, the operation load prediction of the compressor is crucial for monitoring the assembly process. Minor assembly errors may cause a sharp drop in the operating performance of the compressor, thereby affecting the quality of the compressor. Summary of the invention

[0007] The object of the present invention is to provide a method, device, electronic product and medium for predicting the operating load of a compressor to solve the problems existing in the above-mentioned background technology.

[0008] To achieve the above objectives, in a first aspect, the present invention provides a method for predicting the load of a compressor, comprising:

[0009] Acquire historical data from the database as a training set, the historical data including: key variables of compressor equipment;

[0010] Normalize historical data;

[0011] According to the actual problem, a 1DCNN network model is constructed. The constructed 1DCNN network model includes: input layer, convolution layer, fully connected layer, and output layer;

[0012] The data in the training set is input into the 1DCNN network model to obtain the real-time prediction value of the 1DCNN network model for the operation load variable, and the network structure and parameters are optimized according to the difference between the real-time prediction value and the actual value;

[0013] The optimized 1DCNN network model is applied to the online prediction of operation load variables.

[0014] In a second aspect, the present invention provides a compressor load prediction device, comprising:

[0015] The training set acquisition module is used to acquire historical data from the database as a training set, and the historical data includes: key variables of the compressor equipment;

[0016] Normalization module, used to normalize historical data;

[0017] The model building module is used to build a 1DCNN network model according to actual problems. The constructed 1DCNN network model includes: input layer, convolution layer, fully connected layer, and output layer;

[0018] The optimization module is used to input the data in the training set into the 1DCNN network model, obtain the real-time prediction value of the 1DCNN network model for the operation and load variables, and optimize the network structure and parameters according to the difference between the real-time prediction value and the actual value;

[0019] The prediction model is used to apply the optimized 1DCNN network model to the online prediction of operation load variables.

[0020] In a third aspect, the present invention provides an electronic product comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the compressor load prediction method.

[0021] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program implements the compressor load prediction method when executed by a processor.

[0022] The present invention provides an effective solution for compressor assembly process modeling. For the first time, a deep learning algorithm is used to model the compressor assembly process, that is, the compressor operation load is predicted in real time through the assembly process variables, and the compressor performance can be judged based on the load, so as to realize real-time online monitoring of assembly quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a flow chart of a method for predicting compressor operation load provided by an embodiment of the present invention;

[0024] Figure 2 It is an overall architecture diagram of the 1DCNN network model provided by an embodiment of the present invention;

[0025] Figure 3 It is a schematic diagram of the process of offline modeling and online prediction of the 1DCNN network model provided by an embodiment of the present invention;

[0026] Figure 4 is a flow chart of an online prediction process provided by an embodiment of the present invention;

[0027] Figure 5 is a structural diagram of a compressor operation load prediction device provided by an embodiment of the present invention;

[0028] Figure 6 It is a structural diagram of an electronic product provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0029] 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 following embodiments and features in the embodiments can be combined with each other without conflict.

[0030] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0031] like Figure 1 As shown, a method for predicting the load of a compressor in one embodiment of the present invention includes:

[0032] S11, acquiring historical data from a database as a training set, the historical data including: key variables of the compressor equipment.

[0033] S12, normalizing the historical data.

[0034] S13, constructing a 1DCNN network model according to the actual problem, the constructed 1DCNN network model includes: an input layer, a convolutional layer, a fully connected layer, and an output layer.

[0035] S14, inputting the data in the training set into the 1DCNN network model to obtain the real-time prediction value of the 1DCNN network model for the operation load variable, and optimizing the network structure and parameters according to the difference between the real-time prediction value and the actual value.

[0036] S15, applying the optimized 1DCNN network model to the online prediction of the operation load variable.

[0037] The compressor assembly process involves numerous assembly variables. The multivariable complexity of the process makes it difficult to manually monitor and adjust all relevant parameters, increasing the complexity and error of the assembly process, which may lead to a decrease in compressor performance. At the same time, a large amount of data needs to be monitored and analyzed. Traditional assembly processes usually lack real-time detection and feedback mechanisms, so problems may not be discovered until after assembly, and cannot support real-time control and improvement of assembly quality, increasing the cost of rework and repair.

[0038] The present invention provides an effective solution for compressor assembly process modeling. For the first time, a deep learning algorithm is used to model the compressor assembly process, that is, the compressor operation load is predicted in real time through the assembly process variables, and the compressor performance can be judged based on the load, so as to realize real-time online monitoring of assembly quality.

[0039] The mean absolute error (MAE) and mean squared error (MSE) of the 1DCNN model in the prediction task are 5.0624 and 105.984 respectively, which means that its prediction results are closer to the true value and have a smaller error.

[0040] Algorithm design:

[0041] One-Dimensional Convolutional Neural Network (1DCNN) is a deep learning model that is usually used to process one-dimensional data sequences, such as time series, signal processing, and text data in natural language processing. 1DCNN is usually able to learn useful patterns and associations from large-scale data, improving the accuracy and generalization ability of the model.

[0042] 1DCNN has the ability to learn and can adapt to different types of input data. In solving the problem of automatic prediction of compressor performance, there are complex correlations between the variables involved in the assembly data, which may contain complex patterns and features. 1DCNN can automatically learn these features without manually extracting them. 1DCNN can use convolution operations to automatically learn local features in the data, thereby capturing complex features in the data, which is very effective for processing data with spatial correlation or local structure.

[0043] The 1DCNN used in the present invention models the compressor assembly process, such as Figure 2 As shown in the figure, the network contains an input layer, 3 convolutional layers, 1 fully connected layer and 1 output layer.

[0044] (1) Input layer and output layer: The number of neurons in the input layer and output layer of 1DCNN are x and y respectively. The dimension of x is determined by the dimension of the input data, and y is determined by the number of variables to be predicted. The input and output variables involved in the prediction algorithm are shown in Tables 1 and 2. In the present invention, 35 manually measured assembly variables (No. 1-35) and 12 variables automatically collected by sensors (No. 36-47) are considered at the same time to form a 1×47-dimensional vector as input, that is, x=47; 16 compressor performance indicators are considered to form a 1×16-dimensional vector as the output vector, that is, y=16.

[0045] Table 1 Compressor assembly variables

[0046]

[0047]

[0048] Table 2 Compressor performance indicators

[0049]

[0050]

[0051] (2) Convolutional layer: Convolutional layer 1 has 1 input channel and 8 output channels, the convolution kernel size is 3, and the stride is 1. The parameters of convolutional layers 2 and 3 are similar to those of convolutional layer 1, but the number of output channels is 16 and 32 respectively. After each convolutional layer, an activation function is used for nonlinear transformation. The convolutional layer calculation formula is as follows:

[0052]

[0053] Among them, M j Represents the input vector, l represents the lth layer, w is the one-dimensional convolution kernel from the i-th neuron in the l-1th layer to the j-th neuron in the lth layer, b is the bias vector, and the activation function f is the Rectified Linear Unit (ReLU) function.

[0054] (3) Fully connected layer: The output of the convolutional layer is flattened into a one-dimensional vector and input into the fully connected layer, and finally predicted at the output layer. The model captures the features in the data through the convolutional layer, and then makes the final regression prediction through the fully connected layer.

[0055] Prediction model building process:

[0056] 1DCNN is used to model the compressor unloading assembly process, and the compressor operation unloading variables are predicted by the assembly process variables. The overall process is as follows: Figure 3 As shown, it includes offline modeling process and online detection process.

[0057] like Figure 4 As shown, the process of applying the 1DCNN network model to perform parameter prediction in one embodiment of the present invention, that is, applying the optimized 1DCNN network model to the online prediction of the operation load variable, includes:

[0058] S41, collect real-time data at time t and input it into the trained 1DCNN network model.

[0059] S42, the pressure prediction result is output by the 1DCNN network model.

[0060] S43, judging the operating performance of the compressor.

[0061] S44, output the results, and control the assembly process variables according to the result feedback to maintain the reaction stability. At the same time, the input at time t+1 is collected and input into the model.

[0062] S45, measuring the mean absolute error and mean square error of the online prediction results.

[0063] In by Figure 4 In the overall process shown, operation S45 is an optional operation. The other operation steps, that is, the operation steps from S41 to S44, constitute a cyclic process. In actual application scenarios, the online prediction of various operation load variables is completed by cyclic execution of operations S41 to S44.

[0064] By measuring the mean absolute error of the prediction results online, the operation of the model can be judged in real time, and the model can be tuned and troubleshooted in a timely manner.

[0065] like Figure 5 As shown, a compressor operation load prediction device in one embodiment of the present invention includes: a training set acquisition module, a normalization module, a model construction module, an optimization module, and a prediction module.

[0066] The training set acquisition module is used to acquire historical data from a database as a training set, and the historical data includes: key variables of the compressor equipment.

[0067] The normalization module is used to normalize historical data.

[0068] The model building module is used to build a 1DCNN network model based on actual problems. The constructed 1DCNN network model includes: input layer, convolution layer, fully connected layer, and output layer.

[0069] The optimization module is used to input the data in the training set into the 1DCNN network model to obtain the real-time prediction value of the 1DCNN network model for the operating load variable, and optimize the network structure and parameters according to the difference between the real-time prediction value and the actual value.

[0070] The prediction module is used to apply the optimized 1DCNN network model to the online prediction of operation load variables.

[0071] In some embodiments, the input variables of the 1DCNN network model include: cylinder height, cylinder plane parallelism, cylinder groove width, cylinder groove parallelism, lower cylinder head inner diameter, lower cylinder head flatness, and lower cylinder head plane straightness.

[0072] In some embodiments, the output variables of the 1DCNN network model include: outgoing-balanced current, outgoing-balanced flow, outgoing-balanced power, outgoing-pressurization current, and outgoing-pressurization time.

[0073] In some embodiments, convolutional layer 1 has 1 input channel and 8 output channels.

[0074] In some embodiments, the convolution kernel size is 3 and the stride is 1.

[0075] In some implementations, the prediction module includes: an input unit, an output unit, a judgment unit, and a re-input unit.

[0076] The input unit is used to collect real-time data at time t and input it into the trained 1DCNN network model.

[0077] The output unit is used to output the pressure prediction results from the 1DCNN network model.

[0078] The judging unit is used to judge the operating performance of the compressor.

[0079] The input unit is used to output the results again, and the assembly process variables are controlled based on the feedback of the results to maintain the reaction stability. At the same time, the input at time t+1 is collected and input into the model.

[0080] In some implementations, the prediction module further includes: a measurement unit.

[0081] The measurement unit is used to measure the mean absolute error and mean square error of the online prediction results.

[0082] In one embodiment, the present invention provides an electronic product, such as Figure 6 As shown, the electronic device includes at least one processor and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned compressor operation load prediction method.

[0083] Among them, the memory and the processor are connected in a bus manner, and the bus may include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together. The bus can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits through interfaces, which are all well known in the art. The interface provides an interface between the bus and the transceiver, such as a communication interface and a user interface. The transceiver can be one element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices on a transmission medium. The data processed by the processor is transmitted on a wireless medium through an antenna, and further, the antenna also receives data and transmits the data to the processor.

[0084] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory can be used to store data used by the processor when performing operations.

[0085] In one embodiment, the present invention provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the above method embodiment is implemented.

[0086] Those skilled in the art can understand from the above description that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including several instructions for making a device (which can be a single-chip microcomputer, chip, etc.) or a processor (processor) execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes but is not limited to various media that can store program codes, such as USB flash drives, mobile hard disks, magnetic storage devices, and optical storage devices.

[0087] In the several embodiments provided in the present application, it should be understood that the disclosed system, device or method can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of modules / units is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or units can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules or units, which can be electrical, mechanical or other forms.

[0088] The modules / units described as separate components may or may not be physically separated, and the components displayed as modules / units may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules / units may be selected according to actual needs to achieve the purpose of the embodiments of the present application. For example, the functional modules / units in the various embodiments of the present application may be integrated into one processing module, or each module / unit may exist physically separately, or two or more modules / units may be integrated into one module / unit.

[0089] Those of ordinary skill in the art should further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0090] The descriptions of the processes or structures corresponding to the above-mentioned figures have different emphases. For parts that are not described in detail in a certain process or structure, please refer to the relevant descriptions of other processes or structures.

[0091] The above embodiments are merely illustrative of the principles and effects of the present application and are not intended to limit the present application. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed in the present application shall still be covered by the claims of the present application.

Claims

1. A method for predicting the load of a compressor, comprising: Acquire historical data from the database as a training set, the historical data including: key variables of compressor equipment; Normalize historical data; According to the actual problem, a 1DCNN network model is constructed. The constructed 1DCNN network model includes: input layer, convolution layer, fully connected layer, and output layer; The data in the training set is input into the 1DCNN network model to obtain the real-time prediction value of the 1DCNN network model for the operation load variable, and the network structure and parameters are optimized according to the difference between the real-time prediction value and the actual value; The optimized 1DCNN network model is applied to the online prediction of operation load variables.

2. The method according to claim 1, characterized in that The input variables of the 1DCNN network model include: cylinder height, cylinder plane parallelism, cylinder groove width, cylinder groove parallelism, lower cylinder head inner diameter, lower cylinder head flatness, and lower cylinder head plane straightness.

3. The method according to claim 1, characterized in that The output variables of the 1DCNN network model include: load-balance current, load-balance flow, load-balance power, load-pressurization current, and load-pressurization time.

4. The method according to claim 1, characterized in that Convolutional layer 1 has 1 input channel and 8 output channels.

5. The method according to claim 4, characterized in that The convolution kernel size is 3 and the stride is 1.

6. The method according to claim 1, characterized in that The optimized 1DCNN network model is applied to the online prediction of operation load variables, including: Collect real-time data at time t and input it into the trained 1DCNN network model; The 1DCNN network model outputs the pressure prediction results; Determine the operating performance of the compressor; Output the results, and control the assembly process variables based on the results to maintain the reaction stability. At the same time, collect the input at time t+1 and input it into the model.

7. The method according to claim 6, characterized in that The optimized 1DCNN network model is applied to the online prediction of operation load variables, including: The mean absolute error and mean square error of the online prediction results are measured.

8. A compressor load prediction device, comprising: The training set acquisition module is used to acquire historical data from the database as a training set, and the historical data includes: key variables of the compressor equipment; Normalization module, used to normalize historical data; The model building module is used to build a 1DCNN network model according to actual problems. The constructed 1DCNN network model includes: input layer, convolution layer, fully connected layer, and output layer; The optimization module is used to input the data in the training set into the 1DCNN network model, obtain the real-time prediction value of the 1DCNN network model for the operation and load variables, and optimize the network structure and parameters according to the difference between the real-time prediction value and the actual value; The prediction module is used to apply the optimized 1DCNN network model to the online prediction of operation load variables.

9. An electronic product comprising: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the compressor operation load prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, wherein: When the computer program is executed by a processor, the compressor operation load prediction method according to any one of claims 1 to 7 is implemented.

Images