Diagnosis method for leakage fault of hydraulic drive system of injection molding machine
Technical Field
The invention relates to the technical field of injection molding machine fault diagnosis, in particular to a diagnosis method for leakage faults of a hydraulic drive system of an injection molding machine.
Background
The injection molding machine is high polymer material processing equipment, wherein the injection molding machine is driven by oil pressure to be used in large quantity, and the failure of the hydraulic driving system of the injection molding machine brings difficulty to troubleshooting and maintenance due to high failure frequency, strong concealment, unobvious reaction and delayed effect. The method has the advantages that a predictive model is established for the fault type and the fault degree of the injection hydraulic driving system of the injection molding machine, so that the hydraulic fault of the injection molding machine can be diagnosed and early warned in advance, and the conditions of quality reduction, production efficiency reduction and even forced shutdown in the injection molding production process are prevented.
The traditional diagnosis modes of the hydraulic fault of the injection molding machine comprise a traditional manual inspection method, an expert diagnosis system and an intelligent algorithm system. Traditional manual inspection has no predictability and only can be used for post-processing; the expert diagnosis system uses a knowledge base and an inference machine to complete fault positioning and diagnosis, but the knowledge base is updated slowly, and the system lacks generalization; the intelligent algorithm system can predict the failure cause in advance, has adaptability and needs a large amount of test sample data.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a method for diagnosing the leakage fault of the hydraulic drive system of the injection molding machine. The fault early warning method based on the neural network can obviously improve the diagnosis efficiency and accuracy.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a diagnosis method for leakage faults of a hydraulic drive system of an injection molding machine comprises the following steps:
1) simulating a wear leakage fault and collecting parameter data;
a throttling valve is connected in parallel in a hydraulic circuit of the injection molding machine to simulate an abrasion leakage fault, the size of the throttling valve is adjusted, and a sensor is used for collecting parameter data in an injection molding oil way for executing a process action of injection molding; the sensor is an injection molding machine body controller and an external collection device of the injection molding machine; the parameter data is at least one description parameter in the description parameters of the injection molding oil circuit, including main oil circuit pressure, oil pump flow value, injection oil cylinder pressure value and injection displacement value, and the process action is executed into at least one action in the injection molding process (including mold closing, injection table forward movement, injection action, pressure maintaining, plasticizing, cooling, mold opening and product ejection) of the injection molding machine by executing the process action trigger marking time sequence data.
2) Constructing a hydraulic simulation model in the injection molding process, and extracting simulation data;
establishing a hydraulic circuit and a control system simulation model of the injection molding machine by AMEsim/Simulink combination, introducing a leakage model to simulate hydraulic leakage fault, and collecting simulation parameter data in the circuit; the parameter data comprises at least one of the characterizing variables (output pressure, flow rate and injection displacement data).
3) Verifying the consistency of simulation and test data;
and comparing the time-frequency domain curve characteristics of the parameter data obtained by the test and the simulation, evaluating the correlation of the parameter data, and verifying the consistency of the simulation and the test data. Dividing a time subsequence group according to the hydraulic working action of the injection molding machine, counting time domain characteristic parameters (expectation and standard deviation indexes) of the subsequence group, and calculating to obtain a Pearson product moment correlation coefficient of a data curve frequency domain: the time domain characteristic parameter error is less than 10%, the frequency domain correlation coefficient is more than 0.5, and the reliability and consistency of the simulation model are verified.
4) Expanding sample data;
and changing leakage parameters of a pump, a hydraulic cylinder and a valve element in the simulation model, acquiring output pressure, flow and injection displacement data in the simulation loop, and expanding the sample size.
5) Obtaining a numerical prediction model through a machine learning algorithm;
obtaining a numerical prediction model through a machine learning algorithm, wherein the data samples used in machine learning are the time sequence simulation data acquired in the step 4) and the parameter data acquired in the step 1); the machine learning method is a BP neural network, and specifically comprises the following steps:
classifying fault types and leakage degrees of the sample data, wherein the fault types comprise leakage of a pump, an output cylinder and a valve element, and the leakage degrees are classified into normal, mild and severe;
selecting a training set and a verification set to form an input matrix;
setting a target coding matrix according to the classification faults;
and defining a training network by the input matrix and the target matrix, and training to obtain a numerical prediction model.
6) And inputting sensing data and judging a specific fault mode.
And inputting the perception data to a numerical prediction model, and analyzing the fault type and the leakage degree according to a result matrix.
7) Compared with the prior art, the invention has the following obvious prominent substantive characteristics and remarkable technical progress: the method effectively solves the problems of low acquisition efficiency and high cost of training data samples of the hydraulic fault early warning of the existing injection molding machine; and secondly, a method for diagnosing the abrasion leakage fault of the hydraulic element of the injection molding machine is provided, the prediction precision of the algorithm is high, and the generalization is realized.
Drawings
The invention is further illustrated with reference to the following figures and examples.
FIG. 1 is a schematic flow chart of the present invention
FIG. 2 is a schematic diagram of a simulated wear-leak fault of the present invention
FIG. 3 is a schematic diagram of a neural network topology of the present invention
Detailed Description
The method for diagnosing the leakage fault of the hydraulic drive system of the injection molding machine according to the present invention will be further described with reference to the accompanying drawings and preferred embodiments.
The first embodiment is as follows: referring to fig. 1, a method for diagnosing a leakage fault of a hydraulic drive system of an injection molding machine, is characterized in that: the method comprises the following steps:
step 1, simulating a wear leakage fault and collecting parameter data;
step 2, constructing a hydraulic simulation model in the injection molding process, and extracting simulation data;
step 3, verifying the consistency of the simulation and test data;
step 4, expanding sample data;
step 5, obtaining a numerical prediction model through a machine learning algorithm;
and 6, inputting sensing data and judging a specific fault mode.
Example two: this embodiment is substantially the same as the embodiment, and is characterized in that:
step 1, simulating abrasion leakage fault and collecting parameter data
The principle of a method for simulating the abrasion leakage fault by connecting a throttle valve in parallel in a hydraulic loop of an injection molding machine is shown in figure 2, the throttle valve is connected in parallel between an oil inlet and an oil outlet, and the opening degree is used as the basis of the leakage degree; the sensor is an injection molding machine body controller and an external collector of the injection molding machine; the parameter data is description parameters of the injection molding oil circuit, including an injection oil cylinder pressure value and an injection displacement value, and the time sequence data is triggered and marked by executing process actions, wherein the executed process actions are six actions including mold closing, forward movement of an injection table, injection actions, cooling, mold opening, product ejection and the like in the injection molding process of the injection molding machine.
Step 2, constructing a hydraulic simulation model in the injection molding process, and extracting simulation data
A hydraulic circuit and a control system simulation model of the injection molding machine are established through AMEsim/Simulink in a combined mode, hydraulic leakage faults and leakage degrees are simulated by adjusting leakage model parameters, and output pressure, flow and injection displacement data in the circuit are collected.
Step 3, verifying the consistency of the simulation and test data
And comparing the time-frequency domain curve characteristics of the parameter data obtained by the test and the simulation, evaluating the correlation of the parameter data, and verifying the consistency of the simulation and the test data. Dividing a time subsequence group according to the hydraulic working action of the injection molding machine, counting time domain characteristic parameters (expectation and standard deviation indexes) of the subsequence group, and calculating to obtain a Pearson product moment correlation coefficient of a data curve frequency domain:
wherein, X
iAnd Y
iAre frequency domain data of the experiment and simulation respectively,
and
are each X
iAnd Y
iIs expected to
The time domain characteristic parameter error is less than 10%, the frequency domain correlation coefficient is more than 0.5, and the reliability and consistency of the simulation model are verified.
Step 4, expanding sample data
Leakage parameters of the hydraulic cylinder and the valve element are changed in the simulation model, output pressure data in the oil way are collected, data 400 groups of different leakage types and leakage degrees are obtained, and the sample size is expanded.
Step 5, obtaining a numerical prediction model through a machine learning algorithm
The data samples used by machine learning are the time sequence simulation data collected in the step 4) and the parameter data collected in the step 1); the machine learning method is a neural network, a schematic diagram of a network topological structure 3 is shown, a model of the BP neural network is composed of input layer neurons, hidden layer neurons and output layer neurons, and each node is i, j and k.
The method comprises the following specific steps:
classifying fault types and leakage programs of sample data, wherein the fault types comprise leakage faults of a hydraulic output cylinder and a valve element, and the leakage degree is divided into normal, mild and severe;
selecting a training set and a verification set to form an input matrix;
setting a target coding matrix table 1 according to the classification faults;
fault marking
|
Kind of fault
|
Fault vector
|
Z1
|
Normal output cylinder
|
0001
|
Z2
|
Slight leakage of the output cylinder
|
0010
|
Z3
|
Severe leakage of output cylinder
|
0011
|
Z4
|
Valve element normal
|
0100
|
Z5
|
Slight leakage of valve element
|
1000
|
Z6
|
Heavy leakage of valve element
|
1100 |
And defining a training network by the input matrix and the target matrix to obtain a numerical prediction model.
Step 6, inputting the sensing data and judging the specific failure mode
And inputting the perception data to a numerical prediction model, and analyzing the fault type and the leakage degree according to a result matrix.