CN102095953A - On-line detection method for performance of accumulator charger - Google Patents

Abstract

The invention relates to the fields of the on-line detection and evaluation of an accumulator charger and especially provides an on-line detection method for the performance of an accumulator charger. The on-line detection method provided by the invention comprises the following steps: applying the knowledge of artificial intelligence to comprehensively judge the performance of the charger; calculating the voltage stabilizing accuracy, current stabilizing accuracy and current equalizing coefficients of the charger in real time by collecting the current, voltage and environment temperature data of the charger in real time; carrying out real-time alarming on the calculating information exceeding a definite value; inputting the data into an artificial neural network; generating a comprehensive judging value through learning; and providing the performance evaluating result of the charger, thus providing the key reference data for the state examination and maintenance of the charger. The on-line detection method for the performance of the accumulator charger provided by the invention solves the problems that the charger of an electric power system can not realize on-line performance evaluation, can not perform the real-time alarming on charger equipment with low performance or hidden dangers, can not provide effective decision data for realizing the state examination and maintenance of the charger, and the like.

Description

A kind of battery charger performance online test method

Technical field

The present invention relates to the online detection and the evaluation field of battery charger, specifically provide a kind of battery charger performance online test method.

Background technology

Charging set is the visual plant of production and maintenance free cell, is mainly used in the charging of accumulator.The quality of accumulator, performance, serviceable life, extremely important with charging set quality, property relationship.Have a large amount of charging set equipment in the electric system, all the time, the maintenance of charging set all is just to carry out after breaking down, and lacks a kind of online appraisal procedure, estimates the performance of charging set based on the operation duty of charging set, accomplishes repair based on condition of component.

Summary of the invention

The present invention is exactly at above problem, a kind of battery charger performance online test method is provided, it has overcome the electric system charging set can not realize the on-line performance assessment, can not be to performance low or exist the charging set equipment of hidden danger to carry out Real-time Alarm, can not realize that repair based on condition of component provides problems such as effective decision-making data to charging set.

The technical solution adopted in the present invention is as follows:

A kind of battery charger performance online test method may further comprise the steps:

The real-time running data of A, collection charging set;

B, calculate the precision of voltage regulation, precision of steady current and the current stabilizing factor of charging set according to the real-time running data of gathering;

C, the precision of voltage regulation with the charging set that calculates, precision of steady current and current stabilizing factor input artificial neural network save as charging set operational factor table, and make contrast with the data in the established charging set operational factor table, draw charging set performance evaluation result.

The accumulated time of gathering real-time running data in the steps A surpasses half an hour.

Real-time running data comprises voltage, electric current or the environment temperature of charging set.

Steps A also comprises the step that the real-time running data of gathering is provided with the alarm restriction, thereby whether the current operation of real-time judge reaches the alarm border, and provides alarm signal.

Step B specifically comprises:

The real-time running data of the charging set that B1, basis collect is judged the running status of charging set;

B2, calculate respectively according to the running status of charging set, if the running status of charging set is then calculated the precision of voltage regulation, precision of steady current and current stabilizing factor for all filling state; If the running status of charging set is a floating charge state, then calculate the precision of voltage regulation.

A kind of battery charger performance online test method of the present invention, it is with the input as artificial neural network of the parameter of a plurality of reflection charging set performances, multifactorial evaluation through the experts database learning to generate, the performance number etc. of output charging set, and provide important evidence to the repair based on condition of component of charging set with this.

Another characteristics of the present invention are above-mentioned artificial neural network have been designed the function of adaptive learning, when charging set under different charge modes, network will be changed learning sample automatically, relearn training, form a kind of new judge system.The function of this dynamic correction experts database makes system draw charging set results of performance analysis more accurately.

Description of drawings

Fig. 1 is the method flow diagram of a kind of battery charger performance online test method of the present invention;

Fig. 2 is the concrete topology diagram of the used Multi-layered Feedforward Networks of a kind of battery charger performance online test method of the present invention;

Fig. 3 is that the charging set duty of a kind of battery charger performance online test method of the present invention is differentiated process flow diagram.

Embodiment

Below in conjunction with the drawings and specific embodiments a kind of battery charger performance online test method of the present invention is further described.

As shown in Figure 1, a kind of battery charger performance online test method of the present invention, it gathers the real-time running data of charging set by the charging set performance analysis system, as parameters such as voltage, electric current, temperature, and be uploaded to background computer simultaneously, enter the software systems treatment scheme by using input interface.

Gather the charging set real-time running data by the charging set performance analysis system, as charging set voltage, current parameters.(accumulation of 2 frame data is arranged) when the data frame number of uploading reaches requirement, enter charging set running status discrimination model, Fig. 2 promptly is that the charging set running status is differentiated flow process, wherein

Z ₁: all fill state;

Z ₂: floating charge state;

Z ₃: other states.

Specify as follows:

The A of current acquisition unit of error, C10 are 10 hours discharge capacities (unit is Ah) of accumulator.

1), all fills state Z ₁:

Electric current＞(0.01 C10+ current acquisition error);

2), floating charge state Z ₂:

-(current acquisition error)≤electric current≤(0.01 C10+ current acquisition error);

3), other states Z ₃:

Remove behind the above state remaining.

(data accumulation that half an hour is arranged) enters the charging set performance analysis models when the data of uploading reach requirement.

When battery condition is in Z ₁The time, and satisfying and to be arranged during data accumulation half an hour, system is with following (1) formula calculating precision of steady current:

${\mathrm{\δ}}_I=\frac{I_M-I_Z}{I_Z}\×100\%---\left(1\right)$

The system backstage is provided with precision of steady current warning definite value, judges whether to exceed the warning definite value after each the calculating, if exceed, provides warning message immediately.

The precision of steady current data are by the SMV service message transmission of IEC61850 standard.

When calculating precision of steady current, write down the current value of each module simultaneously, current stabilizing factor calculates with following (2) formula in system.

${\mathrm{\δ}}_i=\frac{I_m-I_z}{\mathrm{Ie}}\×100\%---\left(2\right)$

The system backstage is provided with current stabilizing factor warning definite value, judges whether to exceed the warning definite value after each the calculating, if exceed, provides warning message immediately.

The current stabilizing factor data are by the SMV service message transmission of IEC61850 standard.

Replace its original initial value with the up-to-date precision of steady current that obtains, current stabilizing factor and Current Temperatures, the substitution network operations obtains the performance number y of charging set.The preliminary formula that provides according to artificial neural network:

Input: net=δ _Iω ₁+ δ _iω ₂+ t ₀ω ₃

Output: $y=f\left(\mathrm{net}\right)=\frac{1}{1+e^{-\mathrm{net}}}\×100\%$

Wherein: y represents charging set performance number, δ _IThe expression precision of steady current, δ _iThe expression current stabilizing factor, t ₀The expression current temperature value, ω _iExpression network weighted value.

When battery condition is in Z ₂The time, and satisfy data accumulation half an hour is arranged after, system is with following (3) formula calculating precision of voltage regulation.

${\mathrm{\δ}}_U=\frac{U_M-U_Z}{U_Z}\×100\%---\left(3\right)$

The system backstage is provided with precision of voltage regulation warning definite value, judges whether to exceed the warning definite value after each the calculating, if exceed, provides warning message immediately.

Precision of voltage regulation data, warning message transmit by the SMV service message of IEC61850 standard.

Replace its original initial value with the up-to-date precision of voltage regulation that obtains and Current Temperatures, the substitution network operations draws charging set runnability value.The preliminary formula that provides according to artificial neural network:

Input: net=δ _Uω ₁+ t ₀ω ₂

Output: $y=f\left(\mathrm{net}\right)=\frac{1}{1+e^{-\mathrm{net}}}\×100\%$

Wherein: y represents charging set performance number, δ _UThe expression precision of voltage regulation, t ₀The expression current temperature value, ω _iExpression network weighted value.

In sum, a kind of battery charger performance online test method of the present invention, the knowledge of its using artificial intelligence is come multifactorial evaluation charging set performance, electric current, voltage and ambient temperature data by real-time collection charging set, calculate the precision of voltage regulation, precision of steady current and the current stabilizing factor of charging set in real time, and can carry out Real-time Alarm to surpassing constant value calculation information, by above data input artificial neural network, through learning to generate the multifactorial evaluation value, and provide charging set performance evaluation result, for the repair based on condition of component of charging set provides crucial reference data.

Above-described embodiment, the present invention embodiment a kind of more preferably just, the common variation that those skilled in the art carries out in the technical solution of the present invention scope and replacing all should be included in protection scope of the present invention.

Claims (5)

1. battery charger performance online test method may further comprise the steps:

The real-time running data of A, collection charging set;

B, calculate the precision of voltage regulation, precision of steady current and the current stabilizing factor of charging set according to the real-time running data of gathering;

C, the precision of voltage regulation with the charging set that calculates, precision of steady current and current stabilizing factor input artificial neural network save as charging set operational factor table, and make contrast with the data in the established charging set operational factor table, draw charging set performance evaluation result.

2. a kind of battery charger performance online test method according to claim 1 is characterized in that, the accumulated time of gathering real-time running data in the described steps A surpasses half an hour.

3. a kind of battery charger performance online test method according to claim 1 and 2 is characterized in that described real-time running data comprises voltage, electric current or the environment temperature of charging set.

4. according to any described a kind of battery charger performance online test method among the claim 1-2, it is characterized in that, wherein steps A also comprises the step that the real-time running data of gathering is provided with the alarm restriction, thereby whether the current operation of real-time judge reaches the alarm border, and provides alarm signal.

5. according to any described a kind of battery charger performance online test method among the claim 1-2, it is characterized in that wherein step B specifically comprises:

The real-time running data of the charging set that B1, basis collect is judged the running status of charging set;

B2, calculate respectively according to the running status of charging set, if the running status of charging set is then calculated the precision of voltage regulation, precision of steady current and current stabilizing factor for all filling state; If the running status of charging set is a floating charge state, then calculate the precision of voltage regulation.

Images